Reading Time and Words per Minute

A reading-time estimate answers a planning question — roughly how long will this take someone to read? — and the mechanism behind it fits in one line: divide the word count by a reading rate. This guide explains that line: where the 238 words-per-minute rate comes from, what an average can and cannot tell you, how the word count is found, how the result is rounded, and where the estimate breaks. Every figure below was computed by the engine that powers the text analyzer, not typed in by hand.

The formula: word count divided by a reading rate

The engine computes reading minutes = words ÷ 238. Nothing else enters: not sentence length, not vocabulary, not syllables, not paragraph structure. Two texts with the same number of words receive the same reading time whatever they say and however they say it. The estimate is therefore linear — every word costs the same 0.25 seconds, so doubling the words doubles the time — and the only lever that moves it is the word count.

A short original paragraph shows the arithmetic end to end. Take this text: "The community garden opens at seven on Saturdays. Volunteers sort seed packets by planting month, then water the raised beds before the sun gets high. Anyone who stays past ten helps carry compost from the far shed. Bring gloves, a hat, and something to drink, because the morning goes quickly once the work starts." The engine finds 54 words in 4 sentences. Divided by 238, that is 0.227 minutes, or 13.61 seconds before rounding, displayed as "14 sec". The length of the sentences and the difficulty of the words played no part; the word count alone did.

Where 238 words per minute comes from

The rate is the adult silent-reading average from Brysbaert's 2019 meta-analysis of reading-rate studies — a meta-analysis being a study that pools the measurements of many separate studies into one estimate. Two words in that description carry the weight. Silent: the figure describes reading to oneself, not reading aloud, which is why the engine keeps a separate and slower constant for speaking time. Average: the figure sits in the middle of a spread across populations and genres. Skimmers read faster than it; technical material reads slower. The engine applies the one rate to every text and every reader, without adjustment, because it has no information about either.

The result is an estimate for planning, not a stopwatch. Read it as "a typical adult reading silently would take about this long," never as "you will take this long" or "your readers should take this long." An average describes a population; it is not a target for an individual or a norm to hit.

How the word count is found

Because the word count is the only input, every convention in the word counter passes straight through to the reading time. The engine's rule, documented in full in what counts as a word, treats a word as a Unicode-aware run of letters and digits in which internal hyphens and apostrophes do not break the run. So "state-of-the-art" is one word, "it's" is one word, and numbers count as words — but "2,500" is two words, because the comma ends a run. A pasted URL such as https://example.com becomes three words.

Figures make the consequence visible. Take this line: "Order 4182 shipped 2,500 units on 14 March at 3.75 each, split across 4 pallets and 2 trucks." A person sees one order entry; the engine finds 20 words, because "2,500" splits into two and "3.75" into two, and it charges each fragment the same 0.25 seconds as "units" or "and". The display reads "5 sec". Whether a reader lingers on a figure is not something the formula knows; it counts tokens. A sentence carrying a link behaves the same way: "Read the full state-of-the-art review at https://example.com before Friday's meeting." is 12 words, of which the address contributes three, and displays as "under 5 sec". Drafts dense with figures, links, or code therefore carry slightly higher word counts, and slightly longer estimates, than their running prose alone would.

How the estimate is rounded and displayed

The formula returns minutes as a decimal; the display converts that to whole seconds and then chooses one of a few formats. First the total is rounded to the nearest second. A total of zero — an empty box — is shown as "0 sec". A total that rounds to a positive number below five seconds is shown as "under 5 sec" rather than as a figure with false precision. Under a minute, the display shows seconds only; at an exact number of minutes it shows minutes only; otherwise it shows both.

Each branch is easy to reach. The headline "Rain delays the harvest again." is 5 words: 1.26 seconds before rounding, displayed as "under 5 sec". The garden paragraph above lands at "14 sec". A text of 476 words divides to exactly 2 minutes and displays as "2 min", with no seconds shown, while 1,000 words display as "4 min 12 sec".

Rounding produces two effects worth knowing. The first is a threshold. 17 words come to 4.29 seconds and display as "under 5 sec", but 18 words come to 4.54 seconds, round up, and show as "5 sec" — one added word changes the label. The second is a stray second on long texts. 5,000 words come to 1260.504 seconds, which rounds to 1261; the display therefore reads "21 min 1 sec", 1 second past 21 whole minutes. That second is the remainder of a division, not a measurement of anything.

Reading time for common word counts

The table answers the question most people arrive with — how long does a text of a given length take? Each row was produced by building a text of exactly that many words and passing it through the engine's counting and formatting functions, the same calls the analyzer makes on your text. The speaking column uses the 150 wpm constant explained below.

Words Reading time (238 wpm) Speaking time (150 wpm)
100 25 sec 40 sec
250 1 min 3 sec 1 min 40 sec
500 2 min 6 sec 3 min 20 sec
750 3 min 9 sec 5 min
1,000 4 min 12 sec 6 min 40 sec
2,000 8 min 24 sec 13 min 20 sec
5,000 21 min 1 sec 33 min 20 sec
10,000 42 min 1 sec 66 min 40 sec

Read the rows as a scale, not as targets: the table says nothing about how long a text should be, only what the formula returns for a given count. Between rows the relationship is linear.

How many words fit in a given reading time

Planning often runs the other way: a slot of known length, and a question of how many words it holds. Invert the formula — words = minutes × 238. The last column is a check: each word count was fed back through the engine, and the display it produces is shown as the engine formats it.

Reading time Words (minutes × 238) Engine display for that count
1 min 238 1 min
2 min 476 2 min
3 min 714 3 min
5 min 1,190 5 min
10 min 2,380 10 min
15 min 3,570 15 min
30 min 7,140 30 min

The check matters because the display rounds: a count a few words off a row still shows that row's whole minutes if the difference rounds away, and a count between rows shows minutes and seconds. The formula gives the count at which the display is exactly a whole minute; the region showing that label is wider than one word.

Reading time versus speaking time

The engine keeps a second constant, 150 words per minute, a typical prepared-speech pace, and computes speaking time from it in exactly the same way: words ÷ 150. Because both estimates are linear in the same word count, their ratio is fixed. Silent reading at 238 wpm runs 1.59 times as fast as speech at 150 wpm, for every text. The garden paragraph reads in "14 sec" and speaks in "22 sec" — 8 seconds apart on 54 words — and the table's speaking column grows in step with its reading column.

One caution applies more sharply to speech. Number-heavy text runs long when delivered aloud because "2,500" counts as two words yet is voiced as four, so a script full of figures, dates, and prices takes longer to say than its word count suggests. The speaking time calculator and the companion guide on speaking time and speech length cover the spoken side; use them for scripts, and this page's estimate for anything read silently.

Where the estimate breaks

The reader is not average. Someone skimming for a name, or working through technical material, is exactly the reader the population average does not describe, and the estimate cannot know which one it is serving.

The text is not running prose. Everything pasted counts — headings, captions, bylines, footnotes, a reference list, the boilerplate at the bottom of an email. Trim to what a reader will actually read before estimating, or accept that the figure covers the whole document.

The text is very short. Below a handful of words the display collapses to "under 5 sec", and around a single sentence the rounding threshold moves the label by one word. Reading time is a figure for passages and documents, not headlines.

Difficulty is invisible. Flesch Reading Ease and the Flesch–Kincaid grade respond to sentence length and syllables per word; reading time responds to neither. A dense passage and a plain one of equal length receive the same estimate. If difficulty is what you are managing, score it separately: the guides on Flesch Reading Ease and the Flesch–Kincaid grade level explain what those scores respond to. Nothing feeds back from them into the timing.

Other tools disagree. Two tools differ for two reasons, sometimes both: a different rate, or a different counting rule at the margins — hyphenated compounds, numbers, URLs, headings. Differences of a few seconds on long texts are normal. Where an exact figure is contractual, the recipient's counter is the only one that matters.

Frequently asked questions

How long does it take to read 1,000 words?

About 4 min 12 sec at the 238 words-per-minute average this site uses — an estimate for a typical adult reading silently, not a measurement of any particular reader. Read aloud at 150 wpm, the same 1,000 words take about 6 min 40 sec.

How many words is a 5-minute read?

1,190 words: five minutes multiplied by 238 words per minute. Paste a text of that length into the analyzer and the display reads "5 min". Because the display rounds to whole seconds, a count a few words either side shows the same label.

Where does the 238 words-per-minute figure come from?

It is the adult silent-reading average reported by Brysbaert's 2019 meta-analysis of reading-rate studies. It is an average across populations and genres — skimmers read faster and technical material reads slower — so the estimate describes a typical reader, not you or your audience.

Why does another tool give a different reading time for the same text?

Two reasons, sometimes both: the other tool assumes a different reading rate, or it counts words differently at the margins — hyphenated compounds, numbers with commas, URLs, headings. Small differences on long texts are normal. If an exact figure is contractual, the only counter that matters is the recipient’s.

Does harder text get a longer reading time here?

No. The estimate is words divided by 238; sentence length, vocabulary, and syllables never enter it. Two texts of equal word count get the same reading time however difficult they are. Difficulty is measured separately by the readability checker's Flesch scores, and nothing from those scores feeds back into the timing.

Is reading time the same as speaking time?

No. Speaking time divides the same word count by 150 wpm, a typical prepared-speech pace, so it is always longer: silent reading at 238 wpm runs 1.59 times as fast. The garden paragraph on this page reads in 14 sec and speaks in 22 sec.

Every figure in this guide was computed by the engine that runs the tools, which executes locally in your browser — nothing you paste is transmitted or stored. Reading speed is a population average, as documented on the methodology page.