The number you are being compared with
When someone asks whether 45 words a minute is good, they need something to compare it with. The best yardstick I know comes from a study by researchers at Aalto University and the University of Cambridge, presented at the CHI conference in 2018 as “Observations on Typing from 136 Million Keystrokes” (Dhakal, Feit, Kristensson and Oulasvirta). Volunteers copied short sentences in an online test, and the final dataset holds 136,857,600 keystrokes from 168,960 people.
The paper reports an average of 51.56 words per minute, with a standard deviation of 20.2. The fastest people in the data reach 120 words a minute or more. The authors also mark out the two ends: the fastest tenth of participants typed above about 78 words a minute, and the slowest tenth below about 26.
| Your speed | Where it sits in the study |
|---|---|
| Under 26 WPM | The slowest tenth of participants |
| About 52 WPM | The average participant |
| Over 78 WPM | The fastest tenth |
| 120 WPM or more | The very top of the dataset |
The authors point out that their participants were self-selected: people who chose to sit a typing test online. Keep that in mind before you treat the average as everyone’s. It is also an English figure. Urdu and Arabic are counted differently, and I come back to them below.
How this site counts a word
Typing tests disagree partly because each counts a little differently, so here is exactly what our Typing Test does in English.
A word is five characters. That is the standard convention in text-entry research, and the CHI study uses it too: in its words, “one word consists of any five characters”. When you finish a word and press Space, the test compares it with the word it asked for. Only correct words count toward your speed: their letters, plus one for the space after each, divided by five and then by the minutes gone. A word with a slip in it adds nothing.
So two people who type the same number of keys in a minute can get very different scores. Every wrong word is typing time that earned nothing, and the result card counts those words for you.
Accuracy is the share of your keystrokes that matched the text at the moment you pressed them. Backspace itself is never counted as a keystroke, but the wrong letter you typed before deleting it already counted. Fix the slip before you press Space and the word still counts toward your speed. After Space, the word is final.
The clock starts with your first letter, so reading the first line before you begin costs you nothing. Besides words per minute and accuracy, the result shows correct characters a minute (spaces included), the number of correct and wrong words, and your best score for that language and length, which is kept in your browser.
What good accuracy looks like
Speed and accuracy get searched together, and accuracy is the one an employer notices first. In the CHI study, the average participant left 1.167% of the finished text wrong, and 90% of participants left under 2.66%. Most people, in other words, correct most of their mistakes.
That figure measures something different from our accuracy. The study counted errors still in the finished sentence; our accuracy counts every wrong keystroke, including the ones you deleted. A score of 96% here means four of every hundred keys you pressed were wrong when you pressed them, even if you caught every one.
The wrong-word count on the result tells you when to slow down. If it climbs past a handful in a minute, your fingers are ahead of your eyes. Speed built on wrong words does not count on this test, and it will not count in a job test either.
When a job asks for a number
Most searches for a “good typing speed” have a job behind them. The only reliable answer is the figure in that job’s advertisement, along with how the employer tests it.
The clearest published rule I have found is in US federal job advertisements on USAJOBS. A Secretary (Office Automation) posting from the US Army Recruiting Command says: “A qualified typist can type 40 words per minute based on a 5 minute sample with three or fewer errors.” A Clerk (Office Automation) posting from the Air Force Materiel Command asks for “a minimum of 40 WPM proficiency”, lets applicants self-certify it, and warns that anyone who leaves out their typing speed is ineligible.
Look at what that rule actually asks. The sample lasts five minutes, and three errors is the limit for all five. To check yourself against it, take the 5-minute test here and read the wrong-word count alongside the speed. The 10 Minute Typing Test is for exams that run longer, where the question is whether your pace holds after the first few minutes.
Data entry advertisements often quote keystrokes per hour (KPH) instead. Multiply your characters a minute by 60 for a rough hourly figure: 200 characters a minute is about 12,000 an hour. A data entry test made mostly of figures on a number pad measures a different skill from typing words, so a word test gives only a rough idea for those jobs.
For law school, the question comes from exams typed on a laptop against the clock. I have not found a law school that sets a minimum speed. What helps in a timed exam is typing fast enough that your hands never hold up your thinking, and accurately enough that you are not going back to fix things.
Programmers type brackets, symbols and long names that no word test asks for, so their speed on everyday words says little about their speed in an editor. For them the test is most useful as a check on one habit: whether they look down at the keys.
Try it: Typing Test
From 30 seconds to 10 minutes, in English, Urdu or Arabic, scored the way this guide describes. Your best score for each length stays in your browser.
Typing speed in Urdu and Arabic
The five-character word was made for English. Urdu and Arabic letters join, short vowels may or may not be written, and a word’s length on screen says little about the effort of typing it. So our Urdu Typing Test and Arabic Typing Test count whole correct words per minute. A score in Urdu cannot be set against a score in English, and neither can be set against the 51.56 average from the CHI study, which was measured on English sentences.
Two rules keep the count fair across keyboards. An Arabic keyboard types ي ك ه where Urdu writes ی ک ہ, and the test treats each pair as the same letter, so a keyboard is never marked wrong for being the one you own. Vowel marks (tashkeel), the tatweel stretch and the invisible joining characters are ignored when two words are compared.
If you have no Urdu keyboard on your computer, switch on English keys and type by sound: each Urdu letter sits on the English key that sounds like it, with Shift for the second letter of a pair, so t gives ت and Shift with T gives ٹ. A map of the keys shows under the box while you type.
Many people in Pakistan write Urdu in Latin letters in messages (“aap kaise hain”). Turning that into Urdu script is a different job from typing it, and Roman Urdu to Urdu does it for you.
Getting faster without getting sloppy
The CHI study also looked at what separates fast typists from slow ones. Two of its findings are useful in practice.
The first is fingers. Participants reported typing with anything from one or two fingers to all ten, and speed rose with the number of fingers used: fast typists averaged 8.4 fingers, slow typists 5.3. People who had taken a typing course were on average 5 words a minute faster than those who had not, which the authors call a relatively small difference.
The second is rollover: pressing the next key before you have let go of the last one. The authors found that slow typists use almost no rollover, while most fast typists use it for 40 to 70% of their keypresses, and that it is a strong predictor of speed (a correlation of 0.73). You will not notice yourself doing it. It shows up once your hands know where the next letter is without your eyes going to the keyboard.
What I would take from that for practice:
- Keep your eyes on the screen. Looking down breaks the flow that rollover depends on.
- Use all your fingers, even if it makes you slower for a week or two.
- Practise with Sentences rather than Common words once the common words feel easy. Sentences have capitals and punctuation, as real writing does.
- Take the length of test your exam or employer uses, and watch the wrong-word count before the speed.
An employer’s test runs for its full length, so practise at a pace you can hold to the end with few wrong words. A fast first minute that fades by the third will not pass a five-minute test.
Frequently asked questions
What is a good typing speed?
About 52 words a minute is average: that was the mean among 168,960 volunteers in the 2018 CHI study of typing by Aalto University and the University of Cambridge. Above about 78 puts you in the fastest tenth of that group, and under about 26 in the slowest tenth.
Is 40 words per minute a good typing speed?
It is below the study’s average of about 52, and it is the level the US Army and Air Force job advertisements quoted in this guide ask of a qualified typist: 40 words a minute over a five-minute sample with three or fewer errors. Take the 5-minute test here to check yourself against that rule.
What is a good typing accuracy?
Look at the wrong-word count as well as the percentage. Our accuracy counts every wrong keystroke, even ones you deleted, so 96% means four keys in a hundred were wrong when you pressed them. In the CHI study, most people corrected their slips: 90% left under 2.66% of the finished text wrong.
How do I turn words per minute into keystrokes per hour?
Multiply by 5 for characters a minute, then by 60 for an hour. 40 words a minute is 200 characters a minute, or about 12,000 keystrokes an hour. Our result shows characters a minute directly, spaces included, so you only need to multiply that by 60.
Why is my Urdu or Arabic typing speed lower than my English one?
The two are counted in different units. English speed uses the five-character word; the Urdu and Arabic tests count whole correct words. The figures cannot be set side by side, so compare an Urdu score only with other Urdu scores, and an Arabic one with Arabic.
Written by Saad Naeem
AI engineer in Doha who builds every tool on ToolboxHere. The figures in the guides come from the official sources named in them or from the tools themselves. Found a mistake? Tell us and it is corrected.