The Best Robot Vacuum

To share this page via email, fill out the fields below:
Message Sent!
Oops! Please try again

After running more than 30 cleaning cycles over two months with four of the top robot vacuums, we’ve concluded that the iRobot Roomba 650 is the bot that we’d recommend to most people who want an automated helper to keep the floors tidy.

WAIT: February 2, 2017
We’re changing our picks. Our new top pick for most people is the Eufy RoboVac 11. It costs much less than our previous top pick, but delivers similar results. Do not pay more than $220 for this robot. Our new upgrade pick is the iRobot Roomba 960. It’s the best value among any of the full-featured robots that work with a smartphone app and can clean an entire story of most homes (without getting stuck too often). Look for our full update in a few weeks.
Expand Most Recent Updates
December 2, 2016: We’ve begun testing some of the newest bots. It’s too soon to say if we’ll be changing any of our recommendations. This is a popular time of year to buy a robot vacuum, though, and we wanted to let you know which models we’re testing and what we’re looking for in each of them. See the What to look forward to section below.
August 4, 2016: We’ve added long-term test notes for our picks, and we’ve updated our thoughts on new Dyson, Roomba, Samsung, and Anker models.
July 28, 2016: We’ve added our thoughts on the Dyson 360 Eye to the Competition section below.
February 22, 2016: We've added a note on the typical life span of a robot vacuum to the section on who should buy one. We expect them to last around three to five years, which is typical for a midrange vacuum cleaner.
February 10, 2016: After running more than 30 cleaning cycles over two months, we can say that the iRobot Roomba 650 is our main pick. It cleans up more debris than other bots, and it has the best user ratings of any robot vacuum available. It's quieter and more affordable, too. If you have lots of plush carpet and an open floor plan, we recommend the Neato Botvac D80. It's better at getting hair out of medium-length rugs, and it covers large, open spaces more efficiently. It also has a more predictable nav pattern if you crave order.
November 23, 2015: We’re hard at work testing new robots, but here’s a quick update ahead of the holiday shopping season. For most people, we think the iRobot Roomba 650 ($320-400) is still the smart-money pick. It works well in almost any home (exception: black floors) and has an excellent track record for effective and reliable performance. We still need to do some testing, but user reviews suggest that Neato Botvac 80 or D80 ($450) might be a better alternative if your home has a mostly open, uncluttered floor plan, and you have long-haired pets and/or thick carpets. If you have a large home (each floor is larger than 1,000 sq. ft., give or take), it can be worthwhile to get one of the high-end, latest-generation, app-connected bots. We need to spend more time testing them, but so far, the Neato Botvac Connected ($700) looks like a better bet than the iRobot Roomba 980. It's a close fight though! The Neato Connected has fewer navigational hiccups, and seems to cover more ground more efficiently. It's a polished version of older Neato bots, with a much, much longer battery life. The Roomba 980 ($900) adds a camera to help it map big spaces, but the new system is a little wonky.
September 3, 2015: We added a link to CNET's lukewarm review of the Neato D Series. While it still ranks high enough for us to probably include it in our testing this October we were hoping for a machine that improved on some of the limitations of its predecessors, namely noise. However, Neato has also announced a new Wi-Fi-enabled robot vac ($700) called the Neato Botvac Connected, which comes with mobile app connectivity for remote scheduling and operation. It also adds in a new, quieter Eco mode for less noisy operation. We’re hoping to begin testing this October.
May 28, 2015: Neato has just announced an updated line of its floor cleaning robots called the Botvac D Series. The D series will use Neato’s room mapping laser guidance but it comes with a new, sleeker design and an improved brush system, which will reportedly allow it to clean deeper on a wider variety of surfaces while also reducing overall noise on hard surfaces like wood and tile. We welcome that change, since the Neato XV produced a surprising amount of noise for such a small device in our tests last year. The D Series will be available in July and we’ll keep an eye out for the first reviews.
January 6, 2015: Samsung's Powerbot VR9000 is coming to the US this year, though based on its pricing in Europe, we expect it to be a little too expensive for most people. A company called Ecovacs Robotics also showed off a few robot vacs at CES this week. Some of their lower-end models have been available (though hard to find) in the US since mid-2014. The few reviews that exist aren't favorable—navigation seems to be particularly bad, though the self-emptying feature is a positive touch. We'll take a closer look at their high-end models and update when we know more.
September 4, 2014: Dyson's first robot vacuum is finally official; see What to Look Forward to for details and why we're skeptical. Miele's first robot vac has finally gotten reviewed and you can definitely pass on it; see the Competition section.

In most homes, the Roomba 650 cleans up more debris than other bots because it gets stuck less and covers more ground. It’s also quieter and more affordable than other worthwhile vacuums, and it has the best across-the-board user ratings of any robot vacuum, period. This is a particularly great bot if you live in smaller quarters with a disconnected or crowded floor plan and mostly bare floors and short rugs, but it can work pretty well in almost any home. Other robots are more suitable for homes with lots of open space and medium-thick rugs, so you should shop around a bit, but the Roomba 650 should be the first bot you look at.

iRobot Roomba 650
This tried-and-true robot vacuum keeps floors tidy with minimal cost and human effort. In most homes, the Roomba 650 gets stuck less, covers more ground, and cleans better than its rivals. Consider this bot first.

In homes with more space and less clutter, the Neato Botvac D80 is probably a better fit. Its laser-based navigation allows it to follow a precise, methodical pattern through a room, which helps it clean large areas more efficiently than the Roomba 650. It’s also strictly better than the Roomba at cleaning pet and human hair out of medium-pile carpets. The Neato Botvac D80 isn’t our top pick because it’s more likely to have trouble with certain hazards (like chrome finishes or area rugs with tassels), and it costs more than the Roomba 650. But for some people in some homes, it’ll be a better cleaner.

Also Great
Neato Botvac D80
Lots of plush carpet and an open floor plan? The Neato D80 is better at getting hair out of medium-length rugs, and covers bigger, less-crowded spaces more efficiently. Also, some people just prefer its logical, predictable nav pattern.

Generally, you don’t gain much by spending more than $500 or so on a robot vacuum. But a couple of brand-new, high-end bots work with apps that let you start, schedule, or monitor a cleaning cycle from your office or across town or wherever. If you think that feature will come in handy, we have some advice on what to expect and which models to consider.

Table of contents

Who should buy a robot vacuum

Readers Also Like

The Best Broom, Dustpan, and Dust Mop

The Best Broom, Dustpan, and Dust Mop

We spent 125 hours researching and testing 215 brooms, dustpans, and dust mops to find the best.

Read More

You don’t need a robot vacuum. But if you want one, then hell yeah, get one. A robot can do a great job keeping your floors tidy with very little effort on your part. It’s more than just a novelty or a toy. You’ll still need a human-operated vacuum for tough jobs like deep-cleaning dust out of plush carpet, but a bot lets you wait longer between big cleanings, if you want.

We interviewed Sal Cangeloso, who has owned a few iRobot Roomba models and reviewed several robots while he was the editor of Geek. “[Robots] are best at what I’d call maintenance cleaning,” he told us. “The human does the big clean, say, once a month, and then you have the robot clean a few times a week. This’ll keep your place clean and make it so that a few missed corners and stuck-on grime aren’t a big deal.” Wirecutter senior autos editor John Neff uses his Neato XV-21 in a similar way.

You’ll still need a human-operated vacuum for tough jobs … but a bot lets you wait longer between big cleanings.

The raw cleaning abilities of most worthwhile robot vacuums are similar to those of good cordless vacs or very cheap plug-in models: great for debris like tufts of pet hair, crumbs, grit, and anything else that you can feel stuck to the bottom of your feet, but not so great for fine dust that has settled deep into carpet fibers.

Probably the best part of a robot, though, is that it never procrastinates or gets bored. A robot that runs for an hour a few times per week will always cover more ground and pick up more debris than 10 minutes of half-assed, human-operated cleaning a few times per month, even if you’re using the world’s best vacuum.

A robot vacuum will work well in most homes on most kinds of flooring, including hardwood, tile, short carpet, and most area rugs and medium-pile carpet. (Shag carpet, nah, kid.) You don’t need any kind of technical expertise to use one, either. You just press a button to start a cleaning cycle, leave it alone, empty the bin when it’s full, and do a bit of maintenance. You’ll need to rescue your bot from a tangle or other hazard from time to time, and you’ll need to carry it up or down stairs if you want to use it on multiple stories (they weigh between 6 and 10 pounds).

The raw cleaning abilities of most worthwhile robot vacuums are similar to those of good cordless vacs.

And just to set your expectations, robot vacuums are imperfect. They’re designed to adapt to your home and all of its hazards on the fly, and they do that remarkably well considering that every single home is at least a little different. But most people will find that their bot struggles in a particular area of their home—maybe around a certain couch or a corner with a lot of obstacles nearby. The trick is to buy the model that best matches your home and then make adjustments as needed (usually minor) to make the most of your robot helper. A typical life span for a bot is three to five years with good maintenance, which probably seems short given the price. But it’s typical for a midrange vacuum cleaner—and in this case, a robot comes attached to the vac.

And oh yeah, if your dog takes a crap on the floor, the robot can smear it everywhere. Consider yourself warned.


Our finalists, clockwise from top left: Neato Botvac Connected, iRobot Roomba 980, iRobot Roomba 650, Neato Botvac D80.

How we picked

This is the fourth version of our robot vacuum guide. Our last major update was in 2014, and we felt that by the end of 2015, enough had changed in the category for us to take a fresh look. This time around, we put in about 25 hours of research and tested four robot vacuums for about 40 hours over two months—on top of about 100 hours of long-term testing with the iRobot Roomba 650 over two years. (And we’ve continued to use a few models through summer 2016.) We also interviewed representatives from iRobot and Neato and read a couple dozen reviews based on lab tests plus hundreds of user reviews and forum posts.

You can find a lot more robot vacuums these days than you could a few years ago, sold under many more brands. Back in 2014, we tracked down 45 distinct robot vacuums from just 12 different brands. This time, we counted 69 robots from 26 brands. After a decade in which very little changed, the category is suddenly flooded with cheap knockoffs, while prices at the high end continue to rise by hundreds of dollars per year.

We quickly cut any model with no brush roller (it won’t clean carpets), poor availability in the US (that’s a lot of models, actually), or an overall Amazon user rating of less than four stars. After checking out lab tests, user reviews, and forum chatter, we also dismissed any model that cost less than $250 because they’re basically toys with poor suction, small brush rollers, and unreliable navigation. (Since we published this version of the guide, some newer low-cost bots have come out, and they’re actually getting pretty decent. We’re going to reevaluate that criterion for our next big update.)

For our main pick, we decided to test the iRobot Roomba 650 and the Neato Botvac D80 because they had the best specs and the strongest reviews of the 20 bots that passed the initial cut. We also checked out two of the latest-generation robots, the iRobot Roomba 980 and the Neato Botvac Connected, which are the first models in the US that owners can control with a smartphone app. For some people, that could be a worthwhile feature.

How we tested

Our test home is my apartment in Boston. It’s about 900 square feet of living space, including two bedrooms with short knit carpeting, tile in the bathroom, and laminate flooring everywhere else with a few area rugs thrown in. Some areas feel roomy, while others are chopped up a bit with furniture and other fixtures. My wife and I usually take our shoes off in a short entrance hallway to avoid tracking dirt around, but we have a medium-long-haired cat that sheds a lot.

To do some stress testing, we shut the robot into a cluttered room with hazards like stray socks, USB cables, a few chairs, and a low-hanging couch.

During the two-month testing period, we ran about 30 robot cleaning cycles in total and operated each finalist at least six times. We tried to make each session a little different, experimenting with the location of the dock, the cleaning settings, or the brushes when that was relevant, and varying the states of tidiness around the apartment. For the first couple of sessions with each bot, we closely monitored how each bot navigated and how it handled obstacles. The rest of the sessions, we stayed out of the way and let each bot do its thing, checking in with it at the end of the cycle—or when it got stuck, which happened a few times.

To do some stress testing, we shut the robot into a cluttered room with hazards like stray socks, USB cables, a few chairs, and a low-hanging couch. Watching the bots deal with these obstacles gave us a clearer idea of how each bot handled some challenges known to cause problems.

After a couple of years of covering robots, our take is that smart navigation is much more important than raw cleaning performance.

To be clear, we did not run controlled cleaning tests the way that CNETConsumer Reports, and do. Each of those outlets (especially CNET) has some useful data on how well each bot can clean different kinds of debris on different surfaces. We’ll link to their findings throughout this guide whenever they’re relevant. But after a couple of years of covering robots, our take is that smart navigation is much more important than raw cleaning performance. So real-world navigation is where we focused our testing.

Our pick

iRobot Roomba 650
This tried-and-true robot vacuum keeps floors tidy with minimal cost and human effort. In most homes, the Roomba 650 gets stuck less, covers more ground, and cleans better than its rivals. Consider this bot first.

The iRobot Roomba 650 is the smart-money pick for most people who want a robot vacuum cleaner. It’s consistently effective in almost any home while other bots tend to work great in some scenarios but fall short in others. We found that the Roomba 650 is more likely to finish its cleaning cycles on its own without getting stuck or tangled and having to wait for a human to come to the rescue. The nature of its navigation system leads it into tight areas that other bots won’t go near, so it tends to collect more crumbs, pet hair, and other debris overall. Compared with its rivals, it’s quieter, it’s easier and cheaper to maintain and repair, and it’s less expensive up front. For some people, our runner-up will be a better pick, but every would-be bot-vac owner should look at the Roomba 650 first.


The iRobot Roomba 650 can keep the floors tidy in most homes and gets stuck less often than its rivals.

The Roomba 650 has a more adaptive, persistent, and resilient navigation system1 than its competitors, including the Neato Botvac D80. Since it’s so nimble, the Roomba 650 can work well in all kinds of settings—in a crowded floor plan or an open concept, on wood floors or knit carpet, in a single bedroom or covering hundreds of square feet. It might look kind of aimless as it drives headlong into walls, furniture, and other fixtures seemingly at random. As Rich Brown, senior editor at CNET put it to us, the Roomba is “like a party.” And who could forget DJ Roomba?

In our testing experience, though, and judging from the hundreds of user reviews we’ve read, the Roomba 650 doesn’t get stuck or otherwise give up during a cleaning cycle as often as competing models from Neato or other manufacturers. Look at it this way: If you schedule your bot to clean while you’re at work and it gets stuck under the couch 10 minutes into the cleaning cycle, it will sit there all day waiting for you to come rescue it, and your floors will still be dirty when you get home. Defeats the purpose of having an automatic cleaner, no?

The secret is that the Roomba 650 is more adept at escaping “bot traps” such as a rogue USB cable, a maze of furniture legs, the fringe on a rug, or a tall threshold. We think that’s because it relies more on touch-based sensors than other bots do, which gives it a more detailed sense of its immediate surroundings and therefore a clearer escape route. Also, iRobot has been making robot vacuums longer than anybody else, so the company’s software engineers have had more time to fine-tune the algorithms that let the bots escape from potential traps. That makes the Roomba 650 a great navigator in homes with cozy, crowded floor plans. The Neato Botvac D80, by comparison, is better about avoiding obstacles in the first place, but it struggles more when it drives over a cable, bumps into a hazard that its mapping system couldn’t see, or ends up somewhere with no obvious exit. It can also sometimes get stuck on tall thresholds.


The Roomba 650’s freewheeling nav system might look aimless, but it covers a lot of ground during a cleaning cycle and does a great job of escaping or avoiding hazards.

A freewheeling nav system also helps the Roomba 650 cover more ground than its competitors, which gives it a better chance at picking up more debris. notes in a review of the Roomba 650 that by “really ramming itself at tight spaces,” the bot is “able to pick up dirt where other vacuums simply shy away.” We’ve found that it’s more likely than the Neato D80 to drive into tightly clustered groups of tables and chairs, such as the space under a dining-room table. The Neato might just avoid that area altogether if it doesn’t perceive enough space to maneuver freely between the chair legs. So unless you’re willing to do something like put the chairs up on the table, the Neato will pick up none of that debris, and the Roomba will pick up at least some of it.

A longer battery life and a faster cleaning pace also give the Roomba 650 a competitive edge in many homes. It runs for 80 to 90 minutes per cycle, whereas the Neato Botvac D80 squeezes out about 60 minutes and also moves at about half the speed of the Roomba while it’s turning or cleaning the edges of a room. (It moves about as fast as the Roomba in straightaways.) Depending on the size of the space you’re cleaning, the Roomba can make two or even three passes over most of the floor, while the Neato D80 is designed to make only one pass.

Extra speed, a longer run time, and additional passes aren’t necessarily advantages. If a robot can do the same job in less time, great. But based on our experience, user reviews, and test results from outlets like CNET, Consumer Reports, and, we think that in most real-world settings, the Roomba’s speed and persistence allow it to pick up more debris than the Neato D80 can. That won’t be the case for every home. For example, the Roomba 650 can’t clean medium-pile carpets as effectively as the Neato D80, even after multiple passes. And the Roomba’s semirandom navigation means that in larger homes it’s likely to miss a few patches of floor in a given session, so the Neato D80 might be a more satisfying choice in that case. We’ll discuss more of these scenarios in our runner-up section below.


If you have pets, the Roomba 650 can pick up tons of pet hair—more than you think your pet leaves behind.

The Roomba’s inoffensive whirring and seemingly nonsensical cleaning patterns almost make it feel like a pet.

Robot vacuums are all designed to run while humans are out of the house, but if you’re home while your bot is cleaning, we think the Roomba 650 is the most pleasant bot to be around. We measured its typical operating volume at about 59 decibels, which is about as loud as a conversation in a restaurant or office. The Roomba’s inoffensive whirring and seemingly nonsensical cleaning patterns almost make it feel like a pet. Ry Crist at CNET puts it well: “subtle, playful touches of personality can really go a long way — something that iRobot seems to have mastered in the Roomba after several generations of development.” The Neato D80 is louder by comparison, running at 65 dB using its combo brush or 68 dB with the rubber-flap blade brush. Decibels are measured on a logarithmic scale, which means that the Neato can be nearly twice as loud as the Roomba. The Neato also has bigger spikes in its frequency response, which tend to be grating to your ears. It’s particularly loud around 125 hertz, similar to the whoosh of an HVAC unit.

The Roomba 650 makes replacing broken components easy. “Since I’ve had [my 500-series Roomba], I’ve replaced almost all the parts, because it’s cheap, easy, and frankly fun to do so,” Sal Cangeloso told us. A whole new set of brushes and filters currently costs just around $20, while one new brush for the Neato D80 alone costs $40. Other parts get more expensive—a first-party replacement battery currently costs about $60, for example. So as with almost any vacuum, you’ll have to put some money into keeping the Roomba 650 running well, but based on the prices of replacement parts and what we’ve read about its battery, those costs should be lower than with other robot vacuums.

The Roomba 650 has the best user ratings of any robot vacuum of any price.

Finally, the Roomba 650 has the best user ratings of any robot vacuum at any price: Currently it has an overall score of 4.5 stars (out of five) across 2,641 reviews on Amazon. For as long as we’ve been keeping track, it has also been the best-selling robot vacuum on Amazon.

Most of the positive user reviews seem to recognize that the Roomba 650 is a maintenance cleaner, something that lets you put off human-powered cleanings for a few extra days or weeks (if you can stand it) and shaves some time off those sessions while keeping the floors tidier in the meantime. Pet owners find it particularly helpful for keeping fur off the floor. Some owners have had their Roomba 650 for several years, and it’s still running well. As Amazon reviewer N K Maine puts it, “Overall, I love the Roomba. It’s fun to watch, great to help out around the house to clean for us and really saves us time so we can have one less thing to do when it comes to cleaning up the house.”

Flaws but not dealbreakers

If you have black carpet or black matte wood, laminate, or concrete floors, you should probably check out our runner-up, the Neato Botvac D80.

The Roomba 650 doesn’t really work on nonreflective black or very dark brown floors. Black or dark floors that are a little glossy should be okay. According to iRobot, this limitation has to do with the nature of the ledge sensors, which prevent the bot from hurtling down a flight of stairs. One workaround is to tape over the sensors with white paper or something else small and semireflective. We have not tried this trick ourselves, and the potential downside is that the Roomba can then tumble down those aforementioned stairs. If you have black carpet or black matte wood, laminate, or concrete floors, you should probably check out our runner-up, the Neato Botvac D80, instead.

Since its navigation system relies so much on touch (rather than optics) to feel out a room, the Roomba 650 bumps into furniture and walls dozens of times per cleaning cycle. notes that it hits harder than other bots, at about 3.3 pounds of force, which has the potential to knock wobbly objects off of light tables (but probably won’t). A handful of user reviewers have complained that it left marks, sometimes even scratches, on their furniture. We have not noticed scratches or smudges on any of our own chairs or baseboards, and most reviews don’t mention it as a problem. One easy workaround is to stick a little strip of foam rubber onto the bumper.

However, if you have antique or other priceless furniture that you are absolutely not willing to risk damaging, this is not best bot for you. Any bot will run into your stuff sometimes, but our runner-up and upgrade picks don’t do so as often, and when they do, it’s with less force.

Some people just really can’t stand the “aggravating randomness” of the Roomba.

Some people just really can’t stand the “aggravating randomness” of the Roomba, as Wirecutter senior autos editor John Neff put it. He has owned both Roomba and Neato robots, and he greatly prefers the Neato. “I guess I’m a perfectionist-type, logical person in that regard,” he said. “Watching the Roomba drives me mad.” Plenty of people feel the same way—and that’s fair enough! If you know that the Roomba’s semirandom wandering will raise your blood pressure, check out the Neato D80 instead. That said, Sweethome assistant managing editor Manya Susoev was really irritated by the Roomba 650’s pattern when she first tested it, but she got used to it after a couple of weeks (her pets still hated it, though).

Note too that although the Roomba 650 almost always completes an entire cleaning cycle without getting stuck, the unit doesn’t always make it back to the charger. In the last 20 or 30 minutes of a session, when the battery starts to run low, the Roomba 650 begins to look for the dock, which has an infrared beacon. If the bot sees the dock’s signal, it drives over and parks itself on the charging contacts. But if the Roomba can’t find the dock, it just keeps cleaning until the battery is completely out of juice and then stops in place. That second scenario is more likely to happen in bigger homes or in homes with several small rooms rather than fewer large rooms.

The Roomba 650 has undergone lab testing with only one outlet,, and it did not earn a favorable score in that review. At CNET, Roomba bots generally score lower than Neato bots do.

We aren’t worried about those scores.’s robot-vacuum test course is a small, lofted area with a few obstacles, and it doesn’t look much like any real living space. (Disclosure: I worked there for two years, though I left long before the site started reviewing robot vacuums in 2015.) The writers at seem to realize a disparity between their scoring system and real-world results: The language in the Roomba 650 review is generally positive, despite the poor numeric score. CNET’s test courses have no obstacles at all. And we can see some big disparities among the results at, CNET, and Consumer Reports.

Don’t get us wrong: These reviews absolutely have some useful information. They confirm that Neato bots are better at pulling debris out of thicker carpet than lower-end Roomba models. The review of the 650 gave us some great examples of how that bot’s blind persistence helps it cover more ground than its competitors. But this method of controlled lab testing misses the point that with a category like robot vacuums, there’s no mythical “best” model (yet), just different models that work better in different homes.


Hair gets tangled around the Roomba 650’s brushes pretty quickly, even on the bearings. You’ll need to get your hands dirty and cut it away every few weeks.

Other nitpicks with the Roomba 650: Hair gets tangled around the Roomba 650’s bearings pretty easily. This is mostly something to watch out for if you have pets, but long human hair also has an impact. The mess is most obvious on the brush-roller bearings, but it also builds up around the side brush and the front wheel. Cleanup is simple—you just actually need to do it, or else the buildup can impede performance or even start to damage the robot’s gears. This is a problem with pretty much any vacuum with a brush roller, robot or upright or whatever, not just the Roomba.

The interface on the Roomba 650 is, uh, well, not flashy. Setting the time and scheduling cleanings is a lot like using an old digital alarm clock, and the interface looks sort of like a 1990s vision of the future. No screen, no touch controls, no app or remote of any kind. You can certainly find slicker robots out there now, but we can’t point out anything problematic about this system.

The Roomba 650 can’t really deep-clean carpets—no robot can.

The Roomba 650 can’t really deep-clean carpets—no robot can. We took a Dyson V6 Motorhead cordless vacuum (very powerful, but not even the strongest vac out there) to a carpet that four different robots had cleaned in just two days, and we still managed to pull up quite a bit of dust and a few strands of cat hair. We don’t see this as a drawback, and manufacturers are always up-front about the fact that robots are maintenance cleaners. Yes, they do a hell of a job keeping your floors tidy when you use them properly. But you’ll need something a little stronger to deep-clean your carpets.

Long-term test notes

We first tested the Roomba 650 about two years ago. For the first six months or so, I used it a couple of times per week. During that period, I needed to cut away tangles from the brush rollers about twice a month. Every six weeks, I’d pop out anything that spins, including the rollers, the front wheel, and the side brush, to clear away any bunched-up cat hair.

Since then, I’ve used it less frequently (listen, I review a lot of vacuums and have to let my floor get a little dirty to test them), but it still hit the floor once every few weeks. I’d estimate that it’s done about 100 cleaning cycles total, and it has held up well. So far, I’ve replaced the filter only once, though it really needs replacing again soon. But the side brush is still in okay shape, and the battery still seems to hold a full or near-full charge—it can still clean for 80 minutes at a time.

The most obvious change to the Roomba 650 over time is that its body looks a bit banged up. I noticed this after just a few months of service, and now it has even more shallow scrapes and scuffs. However, I haven’t noticed any impact marks or smudges on my furniture that the bot definitely caused. To be fair, I don’t really care if my IKEA stuff gets a little scuffed up, so I’m not paying very close attention, nor does it bother me that the bot itself looks like it’s been busy.

Better for big homes, plush carpet, and people who crave order

Also Great
Neato Botvac D80
Lots of plush carpet and an open floor plan? The Neato D80 is better at getting hair out of medium-length rugs, and covers bigger, less-crowded spaces more efficiently. Also, some people just prefer its logical, predictable nav pattern.

In some homes, for some people, the Neato Botvac D80 will be a better robot than any Roomba model, including the 650. If you have a very big house, wall-to-wall plush carpet, or lots of pets, or if you know you won’t be able to stand watching the Roomba’s semirandom navigation patterns, you might want to get the Neato D80. The D80 looks a hell of a lot smarter than the Roomba 650, and that’s a big deal for some people. Life is chaotic, so if a robot that drives in straight lines will help bring you a small sense of order or peace or whatever, go with the Neato D80.


The Neato Botvac D80 is a better robot for homes with lots of plush carpeting and open spaces, as well as for people who prefer a bot that follows a logical, methodical route.

One advantage to the Neato D80 is that it picks up more pet hair from carpets than the Roomba 650, especially using the bristled combo brush (rather than the rubber-flap blade brush). In CNET’s tests, the Neato D85 (which is functionally identical to the D80) had a huge advantage with pet hair over the Roomba 880 on both short- and medium-pile carpet. We’ve also read several user reviews by people who have lots of carpeting and have owned both a Roomba and a Neato, and they swear that the Neato picks up more debris. Consumer Reports’s results were slightly different, with the Roomba 880 slightly outperforming the “basic” Botvac 80 on carpets and floors, though we’re not sure to what degree. For what it’s worth, in our nonscientific cleaning testing, we’ve found that both the Neato D80 and the Roomba 650 regularly stuff their dust bins with similar volumes of pet hair, though a lot of our testing is on bare flooring, where both models are capable hair cleaners. Overall, we’re pretty confident in saying that if your home is wall-to-wall plush carpet, the Neato will pick up more debris than a Roomba in most cases, even accounting for navigation issues.


The Neato D80 cleans the edges of a room first, then the middle, before moving on to the next room. It’s a much more logical, methodical cleaning pattern than the semirandom route of the Roomba 650—though it doesn’t necessarily add up to tidier floors.

The D80 really shines in larger homes with uncluttered floor plans. It’s better than the Roomba at methodically navigating big, open spaces without missing any patches because it moves in a set of logical, straight lines.

Neato bots (including the D80) use LIDAR, which is essentially a laser rangefinder, to enable a navigation method known as SLAM (simultaneous localization and mapping). It lets the Neato map out walls and obstacles without bumping into them, unlike most Roomba bots. With that map, the Neato plots a course to clean all of the open floor space in that room, starting with the edges and working its way inward, sort of like a Zamboni cleaning a skating rink. It can adjust the course on the fly when it needs to, such as if it runs into some unexpected obstacles (it still has some touch-based sensors, like the Roomba), or if the lasers find floor space it didn’t see the first time. When it’s finished with one room, the bot moves on to the next one and repeats the process. If the battery looks about to run out before it has a chance to clean the entire area, it returns to its dock, recharges for 90 minutes, and then picks up where it left off.

Life is chaotic, so if a robot that drives in straight lines will help bring you a small sense of order or peace or whatever, go with the Neato D80.

So, whether you want to clean 300 or 3,000 square feet, the Neato D80 can do the job—even if it takes the better part of a day for those big spaces. The Roomba 650, on the other hand, struggles with consistency in spaces larger than 1,000 square feet; since it navigates semirandomly, with no clear coverage pattern, it can simply miss some areas. The more space the Roomba has to wander, the less likely it is to be able to cover each room thoroughly before the battery dies after about 90 minutes. Even if you run it a few times per week, it might miss a few spots here and there. The Roomba is still less likely than the Neato to avoid or get stuck in cluttered areas, though, so if your floor plan is large but cramped, that’ll complicate your decision.

Owners like the Neato D80, giving it a score of 4.1 stars (out of five) at Amazon as of this writing. The Roomba 650 earns a rating of 4.5 stars, by comparison, while the older, “basic” Botvac 80 has an overall mark of just 3.7 stars. Now that you’ve read about the bots’ relative strengths in certain kinds of homes, if you think that the Neato D80 (or some other D-series model) is right for you, go ahead and buy it with confidence—it’s another great robot.

The main problem with the Neato is that it still gets stuck in some silly places.

We still don’t think that recommending the Neato D80 as our top pick is the right call, though, because it’s more likely to give some owners some trouble than the Roomba 650. The main problem with the Neato is that it still gets stuck in some silly places. In one of our test sessions, it drove under a futon and just gave up when it couldn’t figure out how to get back out the same way it came in. It also tangles on cables more easily than the Roomba 650. Other users have reported that Neato bots cannot deal with chrome finish—the LIDAR system doesn’t see it properly, which forces the rest of the sensors to try to course-correct. And the Neato can struggle with certain features, like tall-and-wide thresholds, furniture with inclined legs, or other objects that are tall enough to block the path of the bot but too short for the LIDAR turret to see.


Here, the Neato D80 got itself wedged between a toilet and a cabinet. It’s more prone to getting itself stuck in silly situations than Roomba robots.

It’s also more timid about getting close to obstacles. Representatives from Neato told us that the Botvac D series leaves a few millimeters around objects like walls and furniture. That’s so it gets stuck less often, but this approach has the side effect of leaving parts of the floor uncleaned. Since the D80 also makes fewer passes over any given spot, it tends to pick up a bit less debris per session than the Roomba 650 in our experience. The results can vary depending on your flooring type, but if you’re working with a typical mix of bare floors and somewhat short carpets, our educated guess is that the Roomba will usually pick up more stuff—even if it is just through dumb luck.

If you’re probably going to be at home while your robot works, you might find the Neato D80 to be a little loud. Even with its bristled combo brush in, it’s about 6 decibels louder than the Roomba 650. (The blade brush is even louder, especially on bare floors.) Personally, I was uncomfortable being in the same room as the Neato because of the volume. It’s about as loud as a window air conditioner with the fan at top speed.

And of course, there’s the price. We’ve seen the D80 cost anywhere from around $50 to $130 more than the Roomba 650—not cheap. The Neato does have a color LCD, which you might find useful if you regularly change your bot’s cleaning schedule. It also sports a “high performance” filter, which has at best a marginal impact on air quality and cleaning performance but isn’t something you’ll notice.

What about robots you can control with an app?

The latest generation of high-end, flagship robots like the Roomba 980, Neato Botvac Connected, Samsung Powerbot Turbo, and Dyson 360 Eye allow you to control them with smartphone apps, and they have better batteries that let them clean much larger spaces in less time than their cheaper counterparts. So if you have a sprawling floor plan or want the flexibility of app control, you might be interested in one of these top-end models.

That said, these big-ticket bots cost a ton of money and don’t necessarily clean your floors much better than the midrange bots we recommend. Slight advantages, yeah, but worth hundreds of extra dollars? We think most people will be happier to save the cash.

If price isn’t one of your major concerns, however, here’s what you should know:

Having an app-connected robot vacuum is handy, but not life-changing for most people. The best part is that you can start or schedule cleanings from your phone. If you’re at work and you make last-minute plans to Netflix and chill or whatever, and you need to tidy up your place on short notice, you can do that through the app. Pretty neat, though the app still can’t rescue your bot mid-cleaning if it gets stuck.

From what we’ve read and from our hands-on experience, we think the Neato Botvac Connected is the most sensible pick of the bunch, mostly because it costs less than competing Roomba, Dyson, and Samsung models. But the Botvac Connected has some real problems that stop us from recommending it wholeheartedly.

The Botvac Connected is a polished, upgraded version of the same robot that Neato has been perfecting for five years. Most of what we wrote about the D80 (both strengths and weaknesses) applies here, though this model has a better lithium-ion battery and some software tweaks that help it navigate more deftly, run quieter, and cover much more ground on a single charge—we’d estimate about 1,500 square feet. CNET calls it “the best-performing robot vac we’ve tested,” and gave it a perfect 10. The Neato app is more stripped down than its competitors’ apps, and we ran into some occasional trouble trying to control the bot through the app, but it mostly works as advertised. Bonus: You can manually control the bot through the app like an RC car, though you have to be on the same Wi-Fi network (so no ghost-riding while you’re out of the house).

But we’re troubled by the user reviews for the Botvac Connected. At this writing, it has 3.7 stars (out of five) at Amazon across 202 reviews, which is much lower than the scores of other good bots. Some of the low ratings come from people who had never used robot vacuums before and were expecting better performance from this $700 item, or from people whose homes aren’t really suited to Neato’s style. But the concerning reviews are the ones that highlight Neato’s iffy quality control and customer service. When it works, it works well. But the failure rate seems high next to other bots, and you might have some trouble getting in touch with the company.

The iRobot Roomba 980 is the first Roomba that uses a camera-aided navigation system, which allows it to plot and follow a much more logical, efficient cleaning pattern than older models. That, combined with a longer-lasting battery and a stronger motor, helps the 980 thoroughly clean bigger spaces more effectively. It also has the same bump-’n’-run hardware as older Roomba bots, which gives it excellent trap-escaping agility. Its app is the most robust of the smart-bot bunch. The average user rating is great, too, with 4.4 stars out of five across 493 reviews (though Fakespot gives the quality of the reviews a C, suggesting “indications of inauthentic / low-quality reviews”). awarded it a solid 8.5 out of 10.

Unfortunately, it has some hiccups that are pretty frustrating to find in a $900 robot, most of which stem from the added sophistication in the nav system.

One shortcoming is that the Roomba 980 doesn’t really work in dark or even dusky conditions. In early December, we started a cleaning session around 3:30 p.m. Shortly after sunset, around 4:10 p.m. (why do I live in Boston?), the Roomba stopped running and announced an error code. We asked iRobot about this, and reps replied: “The robot is able to go under dark beds and into a dark room. However, all vision-based systems need at least some light and 980 will have a limited range in very low light. The Error 17 is more likely to happen in a crowded area, like if someone runs it in a dining room in the dark or near dark—because the error in the other sensors becomes too great. iRobot’s customer feedback/studies has found that the vast majority of people run their Roomba during the day.” Sure, and it works fine in artificial light, as well. But low light didn’t used to be a problem for Roomba bots.

Another quirk is that the Roomba 980 runs its entire session at whatever angle it’s facing when it starts. If it’s on its dock and parallel to the wall, it’ll run in straight, clean lines, no problem. If it’s a little crooked, however, it cleans the whole house like it’s a little crooked, and the resulting routes are less efficient than they could be. Neato bots, by comparison, can start at any angle and will always run flush with the walls because they always orient themselves to their surroundings.

iRobot just announced a new connected robot, the Roomba 960. It has the same navigation system and app control as the 980, but with a smaller battery and a less-powerful motor. It costs $700, which is still pretty steep. We’ll see if the company has ironed out any of the nav quirks with this model, but most of the caveats should still apply—it’s an expensive gadget with limited capabilities.

The Dyson 360 Eye, which the company announced back in 2014, has finally surfaced with an eye-watering $1,000 price tag. (To be clear, we have not tested this model yet.) It’s basically an older Dyson cordless vacuum built into a self-driving frame—and although we’re sure it’s phenomenal at sucking up dirt, we don’t think that’s as important as the robot itself, and the bot’s design seems flawed. Its height (4.7 inches) is a problem, and its 45-minute battery life leaves something to be desired. Judging from our interpretations of and CNET testing, the camera-based nav system doesn’t seem to be all it’s cracked up to be, although our real-world testing methods differ from those of both sites. It’s impressive that Dyson did such a good job on its first attempt. But you can find plenty of other robot vacuums that will keep your floors tidy for around $700 less than the 360 Eye.

And finally, there’s the Samsung Powerbot Turbo. Since it became widely available only in the first week of December, not many owner reviews have been published, and no editorial outlets (including us) have tested it yet. But based on what we know about Samsung bots, we wouldn’t get too excited about this model. All the bots the company has released so far have been much, much worse than competing Roomba or Neato models. Samsung would have to make a notably major overhaul to its hardware and software for this bot to be worth $1,000. (See our test notes on the Powerbot Essential, which struggled with basic navigation.)

Yes, some of the Samsung VR9350’s flagship features look amazing on paper. One is that you’ll be able to direct its path with a laser pointer, which has the potential to make spot cleaning much easier than with any other bot we’ve seen. Another is that the bot will be able to create and save a map of your home, and you’ll be able to command the bot to clean a specific room by pointing at that map in the app. Neato bots and the Roomba 980 delete their map data 90 minutes after each session ends, possibly out of security concerns, but Samsung will instead save the data for future use. And if you use the Samsung SmartThings smart-home platform, this bot will work with it. We’re not sure whether these new bots will navigate better than their finicky predecessors, which we covered above, but these are the best “smart” ideas we’ve heard of in robots so far. We’ll believe it when we see it, though.

What to look forward to

As of December 2016, we are testing several new models, including the Roomba 650, Roomba 860, Roomba 960, Neato Botvac D3, Eufy RoboVac 11, Ecovacs Deebot M82, and Hoover Quest 700.

We’re trying to answer three main questions with this new round of testing:

  1. Are the cheap bots from Eufy and Ecovacs good enough to undercut our main pick, the Roomba 650, at least for some people? So far, we don’t think so. But lots of our readers have asked us this question, so we’re looking at it closely.
  2. The last couple of Neato bot generations have been good but always priced higher than the Roomba 650. Now that the Neato D3 is about the same price as the Roomba 650, is it the better buy for most people? The latest Neato has a much better feature set, and its navigation system has had years to mature, so this will be a tight race. The Roomba 650 is still a great robot that we’re happy to keep recommending while we answer this question. But our main pick really could be changing for the first time since 2012.
  3. Are any step-up models worth the extra cash? We’ve shied away from recommending any in the past because they were wicked expensive, with only modest advantages (if any) over our main picks. But this year we’re starting to see more features for less money: The Roomba 860, for example, is the most affordable Roomba yet with the tangle-free extractors. And the Roomba 960 does almost everything that the flagship 980 can do (whole-home navigation, app control) for $100 less. Or the Neato D3 might just undercut everything else anyway. We’re running lots of tests, and we’ll update when we know more.

And no, we are not testing the Dyson 360 Eye. It costs way too much, it hasn’t performed well in other editorial outlets’ tests, and it has a poor customer rating at most retailers’ sites.

The competition

iRobot sells a few other Roomba models aside from our top pick and the models we cover in our connected-bot section. The Roomba 860 costs $500 and has better cleaning hardware than its cheaper siblings. CNET’s testing (of the almost identical Roomba 870) showed that the upgraded rollers, filters,2 motor, and dust bin manage to pick up more debris than the 650. It’s a good design, and you might notice a slight cleaning advantage—if you have lots of carpet. Up to you to decide if that’s worth an extra $175.

For $600, the Roomba 880 works with “lighthouses,” which help the bot subdivide a home into individual rooms. If you’re looking for that capability, a Roomba 900-series or Neato model can do it with less fuss.

A few other discontinued Roomba bots are still floating around from the 500, 600, and 700 series. With these, you can expect similar capabilities to the Roomba 650, though some of them may not have a scheduling feature. If you find one of these bots for cheaper than the Roomba 650, go for it.

Neato has a bunch of other bots as well. All the models within the Botvac D series are similar. We chose the D80 because it comes with both a blade brush and a combo brush, and it’s the model that Neato sent us for testing. The D85 is listed as the same product on Neato’s website, so we believe it’s just an alternate SKU for some retailers to have an “exclusive” model. The D75 costs a bit less than the D80 but does not come with the combo brush, so it’ll have more trouble picking up pet hair. No pets? Take your pick between the models.

You’ll find lots of old Neato XV series units kicking around, but we don’t recommend them. The nav systems make them much, much more prone to getting stuck than the newer Botvac models—the algorithms weren’t as refined. These units had big problems with the batteries as well.

The Botvac “basic” series looks similar to the newer Botvac D unit that we recommend, but they have a few key differences. Neato reps told us that the navigation software handles some obstacles differently, and user reviews suggest that the “basic” model is a little more prone to getting stuck. The “basic” kind also uses different brushes, with bearings that wear out faster. And the blade-style brush is exceptionally loud, because unlike the angled flaps on the Botvac D blade brush, these rubber flaps are flat, so they slap the ground like a seal slapping water. CNET gave the “basic” Botvac a better score than the Botvac D series, but we think the scoring was too harsh, as the reviewers didn’t account for the changes to the nav system or durability, and the test results are within a reasonable margin of error. We’d requested the “basic” Botvac 80 for testing for this guide, but Neato told us that those models are discontinued, and the availability seems to be drying up. If you can find one for a good price, it’s your call, but we think the robots in the Botvac D series are worth paying that little extra.

Samsung makes everything, so of course the company makes a few robot vacuums. We did not plan to test any Samsung bots, because the user reviews are mediocre, editorial reviews are mixed, and such bots tend to be expensive next to models from Roomba and Neato. But Samsung offered to send us the Powerbot Essential, so we checked it out. This model got stuck much more often than any bot we’ve tested for this guide, to the point where it was basically unusable in our test home. It never operated for more than 10 minutes without needing a rescue from a bunched-up area rug or from under a bed where the lighting was kind of dim. Plenty of user reviews point out similarly sad navigation errors. You should probably skip this brand until the company demonstrates a better navigation system.

The Miele RX1 Scout has the fancy pedigree of one of the world’s best vacuum companies, but this robot earns poor user reviews. We didn’t test it. The unit appears to be a rebranded, marked-up version of the Yujin iClebo Arte, which also has mediocre ratings. We can think of no good reason to pick either of these over a Roomba or a Neato.

LG Hom-Bot models are not currently sold in the US. But if you do find one, we haven’t heard of a good reason to get one instead of a Roomba or a Neato.

We found a few bots that cost less than yet look similar to the Roomba 650 and considered calling them in for testing, but decided not to:

The BISSELL SmartClean 1605 has navigation struggles, according to user reviews.

The small single brush on the bObi is likely to struggle with debris pickup, and it has a lower user rating than the Roomba 650. The hybrid mop/vac feature has not proven to be useful in similar bots.

The Ecovacs DEEBOT D83’s small single brush is also likely to struggle with debris pickup, and this model received a poor rating at (though it seems like those reviewers tested a lemon). We’ve seen very few user reviews. And the hybrid mop/vac feature has not proven to be useful in similar bots.

It’s hard to find information on the company that makes the Amtidy A325, so getting service or tracking down replacement filters and brushes will be tough.

Beyond that, we encountered dozens of robots from brands like Moneual, Infinuvo, Ecovacs, bObi, Dirt Devil, iTouchless, Techko, Kobot, Yujin, Robo Maxx, P3, 3Q, Lelec, Agda, and CCP, plus one just called “Auto Cleaner Robot” that we dismissed. Most of them are basically toys. Some of them are listed as current products but basically impossible to find in the US. And a few are just very bad robots, based on what we can tell from specs and user ratings. You don’t need to consider these, and even if you get one for super-cheap in a flash deal or something like that, it’ll probably just prove to be a disappointment.

However, lately we have seen a few low-cost bots that are catching our attention. The main one is the Anker RoboVac 10, which got more attention than most cheap bots because Anker has earned a great reputation for its battery packs and charging accessories. We’ve been testing it out, and it’s better than we expected given the low price (usually around $200).

Should you buy the Anker RoboVac 10 instead of a Roomba 650? Our gut reaction is probably not, because it’s too compromised in too many places. The narrow, single-brush intake can’t pick up debris as efficiently as the cheapest Roomba. Even with the multiple passes that its long-lasting battery enables, it leaves behind some larger crumbs that our main pick can scoop up, and probably some smaller dust as well. It also struggles to transition between bare floors and rugs or other slightly elevated areas.

But to its credit, the Anker RoboVac 10’s nav system seems more clever and resilient than we’ve come to expect from other very cheap bots. Amazon user reviews are strong (and FakeSpot suggests that they’re real reviews). We’re going to continue testing the Anker and keeping an eye on other promising new contenders from brands like iLife and Haier, and we will update this guide if our recommendation changes.

Care and maintenance

Unless you keep a truly immaculate, uncluttered home, you’ll need to tidy up ahead of your bot’s cleaning session. Getting my apartment “bot ready” has become second nature. My ritual is to spend two minutes picking up stray laundry and charging cables, moving my cat’s water bowl from the floor to the kitchen counter, and stowing a couple of small, light area rugs that tend to get bunched up under bots. One reviewer stuffs pillows around the sides of her couch, because her bot will get stuck behind the couch if it can squeeze back there. You might decide to put a virtual wall near an area you don’t want cleaned, like your pet’s food bowls. After the first few cycles, you’ll figure out your bot’s pain points and learn to make quick adjustments to your home to help it run more smoothly.

All bots need a little maintenance. After a few sessions, you’ll probably notice long pet or human hair starting to build up around the rollers. Clean it as necessary. You’ll also need to clean the roller bearings, side brush, and pretty much any other external moving part from time to time to prevent mechanical problems. Some sensors might need wiping off on occasion, too. In most homes, we think any decent bot will stay in top shape with less than an hour of work per month. Filters and brushes need replacing to the tune of about $20 per year if you’re diligent, and the bot will eventually need a new battery, which can be costly.

(Photos by Liam McCabe.)


1. Think of “navigation” in the broadest terms—not just the bot’s spatial awareness and mapping technology, but also its routing patterns, its battery management, and its ability to recover from obstacles, tangles, and other bot-stopping traps. Jump back.

2. We don’t think changing the filter in a robot vacuum affects air quality much, because bots are not perfectly sealed, and they have plenty of places besides the exhaust and filter where air (and fine dust) can leak out. Jump back.

To share this page via email, fill out the fields below:
Message Sent!
Oops! Please try again


  1. Sal Cangeloso, Editor-in-chief at, Interview, December 2013
  2. Mike Fortuna, Forum moderator at Robot Reviews, Interview, December 2013
  3. Rich Brown, Executive editor of reviews at CNET, Interview, November 2013
  4. Melissa O'Dea, Product manager at iRobot, Interview, November 2015
  5. Ken Bazydola, Director of product management – Roomba at iRobot, Interview, December 2015
  6. Matt Tenuta, Director of hardware engineering at Neato Robotics, Interview, November 2015
  7. Megan Wollerton, Best vacuum cleaners of 2016, CNET, January 3, 2016
  8. Robot Vacuums, Reviewed
  9. Best robotic vacuums from Consumer Reports' tests, Consumer Reports, May 22, 2015
  10. Roomba and Dark or Black Surfaces, iRobot
  11. Jonathan Chan, Ecovacs Deebot D83 Robot Vacuum Cleaner Review, Reviewed, December 3, 2015
  12. Jonathan Chan, iRobot Roomba 650 Robot Vacuum Cleaner Review, Reviewed, March 27, 2015
  13. Daniel Wroclawski, In Vegas, Samsung Bets Big on the Connected Home, Reviewed, January 15, 2016
  14. N K Maine, Why would you not buy this?, Amazon customer review for iRobot Roomba 650, May 16, 2015
  15. Megan Wollerton, Neato Botvac D85 Robot Vacuum review: A fresh coat of paint does not a new Neato make, CNET, August 24, 2015
  16. Ry Crist, Neato XV Signature Pro review: Is this boring little bot better than the Roomba?, CNET, September 8, 2013
  17. vic7767, New 5XX Foolproof Cliff Sensor Work around, Robot Reviews, January 7, 2008

Originally published: February 10, 2016

We actively moderate the comments section to make it relevant and helpful for our readers, and to stay up to date with our latest picks. You can read our moderation policy FAQ here.