Reading time is the number of words divided by a reading speed in words per minute: minutes = words ÷ WPM. A 1,000-word article at 238 WPM comes to 4.2 minutes, which is 4 min 12 sec, and a “min read” label rounds that up to 5. The rest is detail: which speed to assume, how to round, and what counts as a word.
The formula
There is one division and nothing else.
minutes = words ÷ WPM
Take an 1,800-word article and a speed of 238 WPM. 1,800 ÷ 238 = about 7.563 minutes. To turn the decimal part into seconds, multiply it by 60: 0.563 × 60 is about 34. The calculated reading time, rounded to the nearest second, is 7 min 34 sec.
The usual slip is reading the decimal as seconds. 7.563 minutes is not 7 minutes 56 seconds; decimals of a minute run from 0 to 1, seconds run from 0 to 60. If you would rather skip the conversion, work in seconds from the start: seconds = words × 60 ÷ WPM, so 1,800 × 60 ÷ 238 = 454 seconds, which is the same 7 min 34 sec.
The formula also runs backwards. words = minutes × WPM tells you how much text fits in a fixed slot: ten minutes at 238 WPM is 2,380 words.
What WPM stands for
WPM is words per minute: the total words in a text divided by the minutes someone took to read it. It is an average over the whole text, so it already includes the slow patches, the pauses and the sentences read twice. Our guide to what WPM means in reading goes further into the measure itself.
The formula needs one WPM value, and choosing it is the only judgement call. The reading time calculator offers three presets and a custom field:
- 238 WPM, the research benchmark. Brysbaert (2019) pooled 190 studies with 18,573 participants and found an average silent reading rate of 238 WPM for English non-fiction among adults, and 260 WPM for fiction.
- 180 WPM, careful. A round planning value for dense or unfamiliar material.
- 300 WPM, brisk. A round planning value for easy, familiar text.
Only the first is a published average. The other two are convenient figures near the ends of the range Brysbaert reports as typical for non-fiction, about 175 to 300 WPM. The same paper notes that the often-quoted “300 WPM average” is higher than the evidence supports, which is why 300 sits here as a fast option and not as the default.
None of these is a target. The research benchmark describes a group of English-speaking adults; the other presets are planning assumptions. Individual rates vary widely. See average reading speed for what the research figures do and do not say.
Exact time and the rounded label
A calculation produces two useful outputs, and they answer different questions.
The exact time is the division result expressed in minutes and seconds: 4 min 12 sec. Use it when you are planning your own time or comparing drafts.
The rounded label is what appears at the top of an article as “5 min read”. On this site the rule is: round the exact minutes up to the next whole minute, with a minimum of 1.
Rounding up, not to the nearest minute, is a deliberate choice. At 238 WPM, 1,000 words is 4.2 minutes. Rounding to the nearest minute gives “4 min read”, and a reader who trusted it runs 12 seconds over. Rounding up gives “5 min read”, and the same reader finishes with time to spare. A label that works as a ceiling is more useful than one that is closer on average but short half the time. Whether a page needs a label at all is a separate question, covered in should your blog show a min read label.
Three cases at 238 WPM show how the rule behaves at the edges:
- 714 words is exactly 3 minutes, so the label is “3 min read”. Nothing needs rounding.
- 715 words is one word over and takes a quarter of a second longer. The label becomes “4 min read”. Every label has a cliff like this somewhere, which is the price of whole numbers.
- 90 words takes 23 sec. Rounding up gives “1 min read”, and the minimum of 1 means no label ever says “0 min read”.
So a label is a coarse summary. Two articles both marked “4 min read” at 238 WPM can differ by more than 200 words.
Worked examples
A short product update, read quickly. 640 words at 300 WPM: 640 ÷ 300 = 2.133 minutes, or 2 min 8 sec. Label: 3 min read.
A technical report, read carefully. 2,500 words at 180 WPM: 2,500 ÷ 180 = 13.89 minutes, or 13 min 53 sec. Label: 14 min read. The same report at 238 WPM is 10 min 30 sec, so the choice of speed moved the estimate by more than three minutes.
A chapter of a novel. 4,200 words at the 260 WPM fiction average: 4,200 ÷ 260 = 16.15 minutes, or 16 min 9 sec. Label: 17 min read.
In each case the arithmetic is trivial. The assumption about speed carries the weight.
How the words get counted
Before any division there has to be a word count, and “word” is less obvious than it sounds. Unicode Standard Annex #29, the specification that defines default word boundaries for software, acknowledges that no single definition works for every language and purpose. Tools therefore make choices, and two tools can disagree by a few words on the same text.
The calculator’s method is short:
- Em dashes, en dashes and double hyphens are treated as separators, because writers often set them with no spaces around them.
- The text is split wherever there is whitespace: spaces, tabs, line breaks.
- Each resulting chunk counts as one word if it contains at least one letter or digit.
That produces these results:
| Text | Counted as | Why |
|---|---|---|
don't |
1 word | No whitespace inside; an apostrophe does not split |
well-known, state-of-the-art |
1 word each | An ordinary hyphen keeps a compound together |
3.5, 1,000, 2026 |
1 word each | Digits count; internal punctuation does not split |
finish—it |
2 words | The em dash is a separator |
10–15 |
2 words | So is the en dash |
and/or |
1 word | A slash is not a separator |
A lone —, & or emoji |
0 words | No letter or digit in the chunk |
Run a full sentence through it: “The state-of-the-art model didn’t finish—it took 3.5 hours.” The count is nine: The, state-of-the-art, model, didn’t, finish, it, took, 3.5, hours. A tool that left the em dash alone would say eight.
Languages written without spaces, such as Chinese, Japanese and Thai, cannot be split this way. For those the calculator relies on the browser’s Intl.Segmenter API where it is available (MDN Web Docs: Intl.Segmenter), and the result is an approximation. The full method is on the methodology page.
Counting differences are usually small. If two tools disagree by ten words on a 1,000-word article, the estimates differ by about one percent, or two to three seconds.
Where the estimate stops being reliable
The speed is assumed, not measured. Across the typical non-fiction range of 175 to 300 WPM, the same 1,000 words takes anywhere from 3 min 20 sec to 5 min 43 sec. Our guide to how long 1,000 words take to read shows that spread at eight speeds. If the estimate is for you, replace the assumption with a measurement from the reading speed test.
Difficulty is invisible to the formula. Just & Carpenter (1980) showed that readers spend longer on infrequent words and at points where information has to be pulled together. Two texts with identical word counts can take quite different times, a problem covered in how text difficulty changes reading speed.
Purpose changes the rate. Reading to learn is slower than reading for the gist, and Rayner et al. (2016) describe speed and comprehension as a trade-off. An estimate at 238 WPM assumes ordinary reading for understanding, not study and not skimming.
Only words are counted. Diagrams, tables, code, equations and embedded video take time that a word count cannot see. Any allowance for them is a rule of thumb you add yourself.
The research describes English-language adults. Applying 238 WPM to other languages, to children or to people reading in a second language stretches the figure beyond what it measured.
These gaps are why the calculator shows the exact time alongside the label and lets you change the speed. Most errors people make with these tools come from forgetting one of the points above; common reading time calculator mistakes lists them with fixes.