Updated September 3, 2026 · By Sumbat.T
What Is a Good Typing Speed? The 168,000-Person Answer
The short answer
- Above 50 wpm is above average. 60 wpm puts you ahead of about two thirds of people. The measured mean is 51.56 wpm across 168,000 volunteers and 136 million keystrokes, with a standard deviation of 20.20.
- The “40 wpm is average” figure repeated across page one is roughly ten words too low and is almost never sourced. It appears to descend from typewriter-era studies of professional typists.
- Typing lessons were worth about 5 wpm. Trained typists averaged 54.35 wpm against 49.00 for the untrained, an effect size of just d = 0.27. What actually tracked with speed was finger count and rollover.
- And the ceiling on typing is far below the ceiling on speech. Conversation runs at 196 wpm, close to four times the typing average, which is why BlabbyAI exists: a Windows app and a Chrome extension that type what you say into any text field, free to start and $8.49 a month unlimited.
Almost every article answering this question gives you a number without telling you where it came from. The most common answer is “the average is 40 wpm, so 50 to 60 is good”, and it appears on typing-test sites, hiring-assessment vendors and HR blogs with remarkable consistency and almost no citations. The number is wrong, or at least badly out of date, and the correction is not subtle: the largest study ever conducted on typing found a mean more than ten words per minute higher.
This page uses that study as its spine. In 2018, Vivek Dhakal, Anna Feit, Per Ola Kristensson and Antti Oulasvirta published Observations on Typing from 136 Million Keystrokes at CHI, based on 168,000 volunteers from more than 200 countries. It remains the definitive dataset on how people actually type, and nothing on the first page of results for this question appears to reference it. Everything below is drawn from the paper itself, with the derived figures clearly marked as derived.
51.56 wpm
The mean, across 168,000 people. Standard deviation 20.20, so the spread is wide.
5 wpm
What a formal typing course was worth on average. Effect size d = 0.27, which is small.
4 to 158 wpm
The full observed range. There is no single “good” speed, only a distribution.
The real average is 51.56 wpm, not 40
The paper states it plainly: “The average WPM value for participants is 51.56 (SD = 20.2).” That is the number to quote, and the standard deviation matters as much as the mean. A spread of 20 wpm around an average of 51.56 means that perfectly ordinary people sit anywhere from the low thirties to the low seventies. Anyone telling you that a specific number is “the” good speed is compressing a very wide distribution into a single point.
The shape matters too. The authors note that the distribution “is not a normal-form one; it shows slight positive skewness”, reporting a skewness of 0.513 and a kurtosis of -0.11. In plain terms: the fast tail stretches further out than the slow tail, so the median typist is slightly slower than the mean, at around 50 wpm. The fastest participants reached 120 wpm or more.
Where the 40 wpm figure probably comes from
The typewriter-era research that most modern articles are ultimately echoing measured professional typists at 60 to 75 wpm, using participants who typed for about eleven hours a week on mechanical or electric typewriters. That literature was never about the general population. Somewhere between those studies and today’s blog posts the number became “40 wpm average” without a dataset attached to it. The 2018 study is the first one large enough to answer the question for ordinary computer users, and it says 51.56.
Typing speed by percentile: where your number actually lands
“Good” is a comparison, so the useful form of the answer is a percentile rather than a threshold. The paper publishes the mean, standard deviation, skewness and kurtosis but not a percentile table, so the figures below are derived from those four published moments using a Cornish-Fisher expansion, which is the standard method for recovering quantiles from known moments when the raw data is not to hand.
They are estimates, and worth treating as accurate to a few words per minute rather than exactly. Two sanity checks suggest the fit is reasonable: the derived 99th percentile lands at 104 wpm, consistent with the paper’s note that the fastest typists reach “120 WPM or more”, and the derived median of 50 wpm sits just below the mean, exactly as a positive skew requires.
Where each typing speed falls
Derived from the published mean (51.56 wpm), standard deviation (20.20) and skewness (0.513) of Dhakal et al. (2018).
5th
21 wpm
10th
26 wpm
25th
36 wpm
50th
50 wpm
75th
65 wpm
90th
79 wpm
95th
88 wpm
99th
104 wpm
| Your speed | Percentile | How to read it |
|---|---|---|
| 30 wpm | 15th | Below average |
| 40 wpm | 31st | Slightly below average |
| 50 wpm | 50th | Average |
| 60 wpm | 68th | Good |
| 70 wpm | 82nd | Fast |
| 80 wpm | 91st | Very fast |
| 100 wpm | 99th | Exceptional |
The practical reading: 50 wpm is the middle, 60 wpm is genuinely good and clears essentially every stated job requirement, 80 wpm is rare, and above 100 you are in the top one percent of a 168,000-person sample. If you came here because a typing test gave you a number and you wanted to know whether it was respectable, that table is the whole answer.
Typing lessons bought about 5 words per minute
This is the finding that most deserves to be better known, and it is the one least likely to appear on a page published by a company that sells typing tests. The study compared participants who reported taking a formal typing course against those who did not:
| Group | Mean wpm | Standard deviation |
|---|---|---|
| Took a formal typing course | 54.35 | 20.80 |
| Never took one | 49.00 | 19.73 |
Roughly five words per minute apart, with the paper reporting the effect as d = 0.27, a small effect by any conventional reading. The standard deviations are nearly identical, so training did not even reliably tighten the spread. The authors put it directly: trained typists type “on average, 5 WPM faster than untrained typists, with a similar SD, of about 20 WPM, for a relatively small difference”.
So if formal instruction is not what separates fast typists from slow ones, what does? The study points at two things, and both are mechanical habits rather than curricula.
| Group | Mean wpm | Time between keys | Fingers used |
|---|---|---|---|
| Fast typists | 69.53 | ~120 ms | 8.4 |
| Slow typists | 20.91 | 480+ ms | 5.3 |
- Finger count. Participants used an average of 6.95 fingers, and 47.6% reported using nine or ten. More fingers correlated with more speed (r = 0.38), with fast typists averaging 8.4 against 5.3 for slow typists. The relationship is non-linear, so the gain is not simply proportional.
- Rollover. Pressing the next key before releasing the previous one. The paper calls this “surprisingly prevalent” and it is one of the study’s two headline new findings. The fastest cluster showed a rollover ratio of 38% and also had the lowest error rates in the entire dataset, so speed and accuracy travelled together rather than trading off.
The eight kinds of typist
Rather than splitting people into trained and untrained, the researchers ran unsupervised clustering on keystroke timings and found that most users fall into eight distinct groups of typists that differ in speed, accuracy, hand and finger use, and rollover. The fastest group averaged around 68 wpm with very short gaps between keys (about 160 ms) and the lowest error rates. Two separate groups sat at around 53 wpm with quite different underlying techniques, which is the clearest evidence in the paper that there is more than one way to arrive at an ordinary typing speed.
Your typing test score is not your working speed
Every number discussed so far, including the 51.56 wpm mean, comes from transcription typing: copying text that already exists. That is what a typing test measures, and it is not what you do at work. When you write an email or a report you are composing, which means deciding what to say at the same time as producing it.
The difference is large and it has been measured. A 1999 IBM study by Karat and colleagues put keyboard users at 33 wpm transcribing existing text and only 19 wpm composing their own: a drop of more than 40% in the same people on the same equipment. Nothing about your fingers changed. The bottleneck moved from your hands to your head.
This reframes the original question. If you are asking whether your typing speed is good because you want to produce written work faster, the honest answer is that your test score is measuring the part of the task that is already not the constraint. Which is where the comparison below becomes the interesting one.
Every text-entry method, in the same units
Typing speed only looks like a fixed constraint until you put it next to the alternative. Here is every mode of getting words out of your head, measured in words per minute, with the study behind each figure:
| Method | wpm | Source | Basis |
|---|---|---|---|
| Typing test (transcription) | 51.6 | Dhakal et al. (2018), CHI | 168,000 people, 136M keystrokes |
| Typing, composing your own text | 19 | Karat et al. (1999), IBM | Same users transcribing managed 33 wpm |
| Typing on a phone | 36.2 | Palin et al. (2019), MobileHCI | 37,370 volunteers |
| Dictating into a phone | 153 | Ruan et al. (2016), Stanford | 32 participants, same task both ways |
| Conversation, total elapsed time | 196 | Yuan, Liberman & Cieri (2006) | 2,438 aligned Switchboard conversations |
| Conversation, silences removed | 236 | Yuan, Liberman & Cieri (2006) | Same corpus, net speaking rate |
The shaded rows are speech. Ordinary conversation was measured at 196 wpm across 2,438 word-aligned telephone conversations, and at 236 wpm once the silences are taken out. Against the 51.56 wpm typing mean, that is a ratio of about 3.8 to 1. We covered where those speech figures come from, and why the 150 wpm number everyone quotes is a convention rather than a measurement, in words per minute speaking.
The cleanest comparison is the 2016 Stanford study, because it had the same 32 people enter the same short messages both ways on the same iPhone. Speech came out 2.93 times faster in English, 153 wpm against 52 wpm. Note how closely that 52 wpm matches the 51.56 mean from a completely different study with five thousand times the sample size. When two independent measurements of ordinary typing land within half a word per minute of each other, the number is probably right.
The honest limits of that 3.8x
Dictation does not make finished writing 3.8 times faster, and anyone claiming it does is quoting the entry rate as though it were the whole task. Two things eat the advantage. Composition is slow in either mode, as the IBM study showed. And dictated text needs reviewing: in the Stanford comparison speech produced fewer errors during entry (5.30% against 11.22%) but left slightly more in the final text (1.30% against 0.79%). The real gain is large on anything you already know you want to say, and smaller on text you are still working out.
The other lever
You already produce words at four times your typing speed
BlabbyAI turns speech into text in any app on Windows, or in any browser tab through the Chrome extension. No voice training, no per-minute billing. 60 credits a week free.
If you do want to type faster
The study points at what actually correlates with speed, which is a more useful guide than a drilling schedule. None of this is complicated, and the honest expectation is that most people who type regularly land somewhere between 70 and 100 wpm at their ceiling.
- Use more fingers, but do not obsess over correctness. Finger count correlated with speed at r = 0.38, and fast typists averaged 8.4. The study is clear, though, that formal touch-typing instruction was worth only about 5 wpm, and a separate motion-tracking study it cites found that people not using all their fingers reached speeds comparable to or greater than trained touch typists.
- Let keystrokes overlap. Rollover was one of the two headline findings, and the fastest cluster overlapped 38% of their keystrokes. This is not something you drill directly; it emerges from typing fluently at a pace slightly beyond comfortable.
- Accuracy is not the trade-off you think. The fastest group in the study also had the lowest uncorrected error rates. Across all participants the mean uncorrected error rate was 1.17%. Slowing down to be accurate is largely a beginner’s bargain.
- Then ask whether it is worth it. Going from 50 to 65 wpm is real work for a 30% gain on the part of writing that was never the bottleneck. Speech starts at roughly four times your typing speed on day one.
What dictation actually looks like as a replacement
If the conclusion you have reached is that improving your keystroke rate is a poor return, the alternative is worth understanding concretely rather than as a statistic. BlabbyAI is dictation software: you hold a shortcut, speak, and your words are typed into whatever text field the cursor is already in. It runs as a Windows desktop app that works system-wide, and as a Chrome extension for browser work and for machines where you cannot install software.
Hold the shortcut, speak, release. The text lands wherever the cursor already was.
The part that matters for anyone coming from a typing-speed question is that raw transcription is not the whole feature. Custom modes let you write a plain-language instruction that reshapes what you said before it lands: tidy it into an email, strip the filler, keep it terse. That covers the editing pass that would otherwise eat back some of the speed advantage the Stanford study measured.
A mode is just an instruction in plain English. Speech is the input to it, not a prompt for it.
It works in any application with a text field, which in practice means Word, Gmail, Slack, Teams, Notion, a browser form or an IDE. Pricing is a free tier of 60 credits weekly with no card, and $8.49 a month for unlimited use. There is no per-minute billing, which is the model most transcription services use and the one that punishes you for talking more.
How to measure your own speed properly
If you want a number you can trust, measure the thing you actually do. A standard online typing test gives you a transcription score, which is the number the percentile table above is built for, so run one of those first and look yourself up. Then run a second, more useful test:
- Open a blank document and write a real email you actually owe someone, for exactly five minutes.
- Count the words in what you produced and divide by five. That is your composition rate, and expect it to be far below your test score. The IBM study found a 40% drop.
- The gap between those two numbers is the part of your writing speed that no amount of typing practice will recover, because it was never about your hands.
A caveat on the study worth stating
The 168,000 participants were self-selected volunteers who chose to take an online typing test, which the authors acknowledge does not permit the control of a laboratory study. People who opt into a typing test may well type faster than the general population, which would make 51.56 wpm an overestimate rather than an underestimate. The paper’s defence, which is reasonable, is that large samples yield better estimates of the shape of a distribution than small controlled ones, and that prior online studies have produced results similar to lab experiments. Treat it as the best available number rather than a census.
Frequently asked questions
What is a good typing speed?
Anything above about 50 words per minute is above the middle of the distribution, and 60 wpm puts you ahead of roughly two thirds of people. Those figures come from the largest dataset ever collected on the question: the 2018 CHI paper "Observations on Typing from 136 Million Keystrokes" by Dhakal, Feit, Kristensson and Oulasvirta, which recorded 168,000 volunteers from more than 200 countries. The mean was 51.56 wpm with a standard deviation of 20.20. Most articles answer this question with "40 wpm is average" and give no source; that number appears to descend from mid-century typewriter studies of professional typists and does not match what modern keyboard users actually do. The honest framing is that there is no single good speed, because the spread is enormous: participants in the study ranged from 4 wpm to 158 wpm. A speed is good if it is fast enough that typing is not the thing slowing your work down.
Is 60 wpm a good typing speed?
Yes. Against the measured distribution of 168,000 people, 60 wpm sits at roughly the 68th percentile, so you type faster than about two thirds of people. It is comfortably above the 51.56 wpm mean and it clears the bar for essentially every job that lists a typing requirement, including most administrative and customer-support roles that ask for 50 to 60 wpm. Worth knowing: 60 wpm is also about the speed the older typewriter-era literature reported for trained professional typists, who were measured at around 60 to 75 wpm. Reaching that on a modern keyboard without formal training is normal rather than exceptional, which tells you something about how much the equipment changed.
Is 40 wpm a good typing speed?
It is slightly below average, at roughly the 31st percentile, though it is perfectly serviceable for most work. The reason 40 wpm gets quoted everywhere as "the average" is unclear and largely uncited; the actual measured mean across 168,000 people is 51.56 wpm, more than ten words higher. If you type at 40 wpm and it is not causing you a problem, there is no reason to treat it as a deficiency. If you do want more throughput, the research suggests your time is better spent on something other than typing drills, because the study found formal typing training was worth only about 5 wpm on average.
What is the average typing speed?
51.56 words per minute, with a standard deviation of 20.20, measured across 168,000 volunteers producing more than 136 million keystrokes in the 2018 Dhakal et al. study. The distribution is not a normal bell curve: it has a slight positive skew of 0.513, meaning the tail of fast typists stretches further to the right than the slow tail does to the left. Practically, that means the median is a little below the mean, around 50 wpm, and a small number of very fast people pull the average up. The study also notes the fastest participants reached 120 wpm or more. For phones the comparable figure is much lower: a 2019 study of 37,370 people measured 36.2 wpm on mobile keyboards.
Do typing lessons actually make you faster?
Much less than the typing-test industry implies. In the Dhakal study, participants who reported having taken a formal typing course averaged 54.35 wpm against 49.00 wpm for those who had not. That is a difference of about 5 wpm, and the authors report the effect size as d = 0.27, which is small. The two groups had almost identical standard deviations of around 20 wpm, so training did not even reliably narrow the spread. What did correlate with speed was the number of fingers used (r = 0.38) and rollover, the habit of pressing the next key before releasing the previous one. Fast typists averaged 8.4 fingers against 5.3 for slow typists. The uncomfortable implication for anyone selling a typing course is that the untrained typists in this dataset had largely converged on the same performance as the trained ones by simply typing a lot.
What typing speed do employers require?
Where a requirement is stated at all, it is usually 40 to 60 wpm for general administrative work, and higher for roles where typing is the job itself, such as transcription or data entry. It is worth being clear about what those thresholds represent: they are hiring conventions, not findings, and the sources quoting them are typically assessment vendors rather than studies. Against the real distribution, a 50 wpm requirement excludes roughly the bottom half of typists and a 60 wpm requirement roughly the bottom two thirds. If you are preparing for a test with a stated threshold, practising the specific test format is more useful than general speed training, because transcription typing (copying text that already exists) is measurably faster than composing your own text.
Why does my real typing feel slower than my typing test score?
Because a typing test measures transcription and your actual work is composition, and those are different tasks with different speeds. In a typing test the words already exist and you are performing a motor task. When you write an email you are deciding what to say while you type, and the thinking is the bottleneck rather than the fingers. A 1999 IBM study measured this directly: keyboard users managed 33 wpm transcribing existing text but only 19 wpm composing their own, a drop of more than 40%. So a 70 wpm test score does not mean you produce 70 wpm of email. This is also why raw typing speed matters less for most knowledge work than the tests imply, and why the useful question is usually about total time to finished text rather than peak keystroke rate.
How fast can you actually get with practice?
The realistic ceiling for most people who keep typing regularly is somewhere in the 70 to 100 wpm range, and the study data supports treating 120 wpm as genuinely exceptional rather than a target. In the 168,000-person dataset the fastest cluster of typists averaged around 68 wpm, and individuals reaching 120 wpm or more were rare enough that the paper calls them out specifically. Two behaviours separate the fast group and both are learnable: using more fingers, and rollover. The fastest cluster showed a rollover ratio of 38%, meaning more than a third of their keystrokes overlapped with the previous one, and they also had the lowest error rates in the study. Speed and accuracy went together rather than trading off.
Is typing speed still worth improving?
It depends entirely on whether typing is what is actually slowing you down, and for a lot of people it is not. If you are at 30 wpm and writing all day, the gain from reaching 55 is real. If you are already at 65 wpm, another 10 wpm is a modest return on a lot of drilling, especially since composition speed rather than motor speed governs most writing. There is also a different lever available now that did not exist when the typing-speed conventions were set: speech. Conversational speech has been measured at 196 words per minute, which is roughly four times the average typing speed, and dictation software converts that directly into text. For anyone whose honest answer is "I want to get words down faster", that is a larger step than any achievable improvement in keystroke rate.
How does speaking compare to typing, in the same units?
Speech is roughly three to four times faster. A 2006 analysis of 2,438 word-aligned telephone conversations from the Switchboard corpus measured an average of 196 wpm counting total elapsed time, rising to 236 wpm once silences are removed. Against the 51.56 wpm typing mean that is a ratio of about 3.8 to 1. The cleanest single comparison comes from a 2016 Stanford study that had the same 32 participants enter the same short messages both ways on an iPhone: speech came out 2.93 times faster in English, 153 wpm against 52 wpm. The gap narrows in practice because dictated text still needs reviewing and because composition is slow in either mode, but the direction and rough size of the advantage are consistent across every study that has measured both. A dictation tool such as BlabbyAI is how that advantage gets used day to day, since it types your speech into any app rather than into a separate transcription window.
Sources
- Dhakal, V., Feit, A., Kristensson, P.O. and Oulasvirta, A. (2018). Observations on Typing from 136 Million Keystrokes. CHI 2018. 168,000 participants, mean 51.56 wpm (SD 20.20), skewness 0.513, kurtosis -0.11. Source of every typing figure on this page.
- Yuan, J., Liberman, M. and Cieri, C. (2006). Towards an Integrated Understanding of Speaking Rate in Conversation. Interspeech. 2,438 word-aligned Switchboard conversations, 196 wpm gross and 236 wpm net.
- Ruan, S., Wobbrock, J.O., Liou, K., Ng, A. and Landay, J. (2016). Speech Is 3x Faster than Typing for English and Mandarin Text Entry on Mobile Devices. Stanford. 32 participants, 2.93x in English.
- Karat, C-M., Halverson, C., Horn, D. and Karat, J. (1999). Patterns of Entry and Correction in Large Vocabulary Continuous Speech Recognition Systems. CHI. Keyboard transcription 33 wpm, composition 19 wpm.
- Palin, K., Feit, A., Kim, S., Kristensson, P.O. and Oulasvirta, A. (2019). How do People Type on Mobile Devices? MobileHCI. 37,370 volunteers, 36.2 wpm.
Your typing speed has a ceiling. Your speaking speed already cleared it.
BlabbyAI is the part that turns speech into typed text: a shortcut on Windows, or the Chrome extension anywhere you cannot install software. 60 credits a week free, $8.49 a month unlimited.