AbilityBenchReading rate
Reading speed test with the comprehension check attached
Read one passage, press Enter when you reach the end, and answer five questions with the passage withdrawn — free, no account, nothing held back, about three minutes. The result gives your words per minute, how many questions came back right, and the two multiplied together, because a rate reported on its own can be raised to any number you like by moving your eyes faster. The reference beside it is 238 wpm for silent non-fiction and 260 wpm for fiction, both from the same pooled analysis.
- 100% free
- No signup
- 285 or 273 words
- 5 questions, passage hidden
- 238 wpm reference
One passage of 285 words, timed from the moment it appears to the moment you say you have finished, then five questions with the passage taken away. The rate and the questions are reported side by side, and neither is reported without the other.
The run, in full
- Stimulus
- one passage of about 285 words, shown in full with no timer on screen
- Your job
- read it once, at a pace you could answer questions about afterwards
- Clock
- starts at the frame that paints the passage, stops at the timestamp on Enter or on the button
- Questions
- five, four options each, passage withdrawn, no going back
- Void
- a read closed inside 8 s, and anything the tab lost the foreground during
- Reported
- words per minute, questions right, and the two multiplied together
Read at the pace you would use for something you intended to remember. There is no way to practice this one without spending the passage, so the honest version of a warm-up is to run it twice and keep the second number.
How to measure your reading speed properly
One passage, one keypress, and five questions that decide whether the number counts.
Choose the report or the story
The non-fiction passage runs 285 words and is compared against the pooled figure for silent non-fiction; the narrative runs 273 words and is compared against the fiction figure, which is about twenty words a minute higher. Pick before you start, because the difference between the two genres is larger than most of what this test could otherwise tell you apart. Neither passage assumes anything you would need to have read before.
Read once, at the pace you would use to remember something
The clock starts on the frame that puts the text on screen and stops on the timestamp the browser attaches to your Enter key, so there is no start button to fumble and no countdown in the corner to race. Read it the way you would read a page you intended to talk about afterwards. Skimming is allowed and will show up: the questions are the thing that tells the difference between reading fast and not reading.
Answer five questions with the passage gone
The text is withdrawn before the first question, so what is being scored is what you took in rather than how quickly you can find a phrase again. Four options each, keys 1 to 4 or the on-screen buttons, no going back. The result panel then shows the rate, the score, the rate multiplied by the proportion correct, and the published average beside all three.
Technical specifications
| Passages | Two, both written for this page: 285 words of non-fiction about the nineteenth-century ice trade, and 273 words of narrative. Word counts are whitespace-delimited tokens counted from the same string that is rendered, so the divisor and the text can never drift apart |
|---|---|
| How the interval is taken | From the paint of the passage to the timestamp on the keypress, both read off the browser's monotonic clock. No countdown, no start button of its own, and no interval measured from a timer callback |
| When a rate is not reported | A passage closed inside 8 seconds — over 2,000 words a minute on either text — is treated as a press rather than a read, and so is any read during which this tab went to the background. The five questions still run and are still scored in both cases |
| Questions | Five per run, four options each, drawn in a seeded order and asked with the passage removed. Guessing all five right has a probability of about 0.001, and the panel prints the exact one-sided binomial figure for whatever you scored |
| Reference figures | 238 wpm silent non-fiction and 260 wpm silent fiction, 190 studies pooled — Brysbaert (2019), How many words do we read per minute? A review and meta-analysis of reading rate, Journal of Memory and Language |
| Rate adjusted for comprehension | Words per minute multiplied by the proportion of questions answered correctly. This page defines that arithmetic on the page rather than citing it, both inputs are printed so you can undo it, and it is not a published index |
| Timing precision | One display frame of quantization at each end — about 17 ms on a 60 Hz panel — against a read that lasts tens of seconds. That is under a tenth of one percent, which is why the rate is a whole number of words and why no part of this page reports a time in milliseconds |
| What leaves the page | Nothing at all during a run. The copy button writes plain text to your own clipboard, including a short token carrying the seed and the figures so the same run can be rebuilt from it |
Frequently asked questions
Is the average adult reading speed really 250 words a minute?
The best available figure for silent non-fiction is 238 wpm, not 250, and the round numbers in circulation mostly have no paper behind them. Brysbaert (2019), How many words do we read per minute? A review and meta-analysis of reading rate, Journal of Memory and Language pooled 190 studies pooled and separated the cases that get blurred together: silent non-fiction at 238, silent fiction at 260 wpm, and reading aloud at 183 wpm, which is a quarter slower because articulation rather than comprehension sets the pace. The 250, 300 and "one page a minute" figures you meet on speed-reading sites are usually quoted with no source at all, and where they are sourced it is to another site quoting a third.
Why does the passage disappear before the questions?
Because a check you can scroll back to answers a different question than the one being asked. With the text on screen, five multiple-choice items measure how fast you can locate a phrase, which is a real skill and not the one a reading rate is supposed to be conditioned on. Taking it away means the items are answerable only from what the read actually deposited. The cost is that a little memory gets mixed into the score, which is why the questions ask about the load-bearing facts of each paragraph rather than about a date in a subordinate clause.
My rate here is lower than on other reading speed tests. Why?
Most of them do not check whether you understood anything, and a rate with no check attached drifts upward for free. The other three causes are visible in the method: the clock here starts at the paint of the text rather than at a button you press when you are ready, so the second or two of settling in is charged to you; the passages are prose with an argument in them rather than sentences assembled to be easy; and nothing on screen is counting, so there is no rhythm to chase. If you want the two numbers separated, run it again and read deliberately slowly — the rate will fall and the score should not.
Can I train myself to read at a thousand words a minute?
No, and the reason is mechanical rather than motivational. Rayner, Schotter, Masson, Potter & Treiman (2016), So much to read, so little time: How do we read, and can speed reading help?, Psychological Science in the Public Interest reviewed the evidence and put the point where comprehension starts to give way at 300-500 wpm. The ceiling comes from the eye and the language system, not from the technique: the eye moves in jumps of 7-9 characters and rests 200-250 ms at each landing, and about 10-15% of those movements go backwards to re-read. Multiply that out and a few hundred words a minute is what the machinery permits. What speed-reading courses teach is skimming, which is genuinely useful, and which fills the gaps from what you already knew rather than from the page.
Does it change the number if I already know the subject?
Yes, substantially, and it is the largest uncontrolled variable in any browser reading test including this one. Familiar material is read faster because prediction does part of the work, and it is comprehended better for the same reason, so prior knowledge moves the rate and the score in the same direction and cannot be subtracted out of either. The passages here are deliberately about things most people have no stake in, and the questions are answerable purely from the text — but if the ice trade happens to be your subject, take the narrative instead and say so to yourself when you read the result.
Why is there no percentile beside my words per minute?
Because the reference figure is a mean with no individual-level spread published behind it, and a percentile needs both. A mean tells you which side of the average you are on; converting it into a rank would require a standard deviation nobody has published for browser-administered reading, and inventing one would produce a confident number about you that no one could check. So this page says above or below, by how much, and stops there.
Does reading on a screen make me slower than reading on paper?
The honest answer is that this page cannot tell you, and neither can any single run of it. Screen-versus-paper comparisons in the literature find small effects that move around with the display, the typography, the length of the text and how the comprehension was checked, and the effect is nowhere near large enough to be visible in one 285-word passage read once. What this page controls is its own typography — one column, generous line spacing, no reflow while you read — so that two runs of it are comparable with each other even if neither is comparable with a book.
Where 238 words a minute comes from, and what the eye can actually do
The number this page compares you against has an author, a year and a method, which is more than can be said for most of the figures in this corner of the web. Brysbaert (2019), How many words do we read per minute? A review and meta-analysis of reading rate, Journal of Memory and Language went through the published rate studies and pooled 190 studies pooled, and the review does the thing that makes the result usable: it keeps apart the conditions that get quoted as though they were one. Silent reading of non-fiction for comprehension came out at 238 wpm. Fiction, read silently, came out at 260 wpm — faster, because narrative is more predictable word by word. Reading aloud came out at 183 wpm, roughly a quarter slower than silent reading, because the mouth becomes the bottleneck. Somebody who times themselves reading to a child and then compares the result with a figure for silent reading has measured the difference between two tasks and read it as a fact about themselves.
The ceiling is worth doing the arithmetic on, because it is the part that makes the speed-reading claims collapse without any need to argue about them. Reading is not a smooth sweep: the eye jumps, lands, and stays put while the text at that point is taken in. Those jumps run 7-9 characters and the pauses run 200-250 ms, per Rayner (1998), Eye movements in reading and information processing: 20 years of research, Psychological Bulletin, and about 10-15% of the movements go backwards over material already passed. Four or five landings a second, each worth a word and a bit, with one movement in eight spent going back, comes out at a few hundred words a minute — which is exactly the band Rayner, Schotter, Masson, Potter & Treiman (2016), So much to read, so little time: How do we read, and can speed reading help?, Psychological Science in the Public Interest identifies as 300-500 wpm, the region where comprehension begins to fall away. The regressions are the detail most often thrown away by advice: they are not a bad habit, they are the process repairing itself, and suppressing them removes something reading needs.
None of that makes a rate worthless — it makes a rate without a check worthless, which is why this page will not print one on its own. Five questions is a thin check and the page says so beside the score; if the accuracy side is what you actually came for, the comprehension test asks twelve questions of four separable kinds and leaves the text in front of you, and the reading level test climbs a ladder of harder and harder passages until accuracy gives way, which answers a different question again: not how fast, but how difficult. Rate also depends on how much of the text you recognize without effort, so if a passage felt slow because of the words rather than the ideas, the vocabulary test and the synonym test measure that directly.
A reaction time here is the interval between the frame that painted the stimulus and the timestamp the browser attached to your key, both read from the same monotonic clock. What neither can see is the display pipeline behind it, so on a 60 Hz screen roughly 16 ms of every figure below is the machine rather than you. That is the timing floor: two numbers closer together than that are the same number, and this page reports no precision it cannot support.
On this page that floor is unusually generous to the measurement rather than threatening to it. The interval being timed here is a read of roughly a minute, not a reaction of a fraction of a second, so a frame of uncertainty at each end is under a tenth of one percent of the result. That is why the rate is given as a whole number of words a minute and why nothing on this page is reported in milliseconds at all: the precision exists, and claiming it would suggest the passage, the mood and the subject matter were controlled to the same standard, which they were not.
This is a measurement exercise, not a clinical assessment. It reports what you did on this page against a stated reference and nothing more — it cannot establish a reading difficulty, a vision problem or a language disorder. Only a qualified professional, working with more than a browser, can make that judgment.
Where your reading time is worked out
Every number on this page is worked out by JavaScript running in the tab you are reading it in. Your answers, your reaction times and your score are never uploaded, logged or kept — which is also why the test carries on working after you disconnect from the network, and why nothing here can be held back behind an email address.
The one timestamp that matters — the paint of the passage — and the one that ends it live in a JavaScript variable in this tab, and the subtraction happens a few lines away from where they were taken. Neither the interval nor your five answers is written to storage, so reloading the page loses a finished run, which is the reason the copy button is there.