How to Measure Your Reading Speed at Home: A Protocol That Gives You a Real Number

Same Person, Two Numbers, Both "Correct"

Pick up a novel one evening, read for ten minutes, do the math: 340 words per minute. The next day you grab your statistics textbook, same procedure, same stopwatch: 180. You didn't lose anything overnight. You measured two different things.

That's the trap of home-made tests. A stopwatch never refuses to produce a number, so you'll always get a figure. Whether it means anything, though, gets decided before you hit start — in choices you don't even register as part of the test.

What counts as "normal" is a separate question, and it has a measured answer: 238 words per minute for non-fiction, from a meta-analysis of 190 studies. This piece is about something else — how you get a number of your own that holds up under a second check, and that you can compare six weeks from now without fooling yourself.

The Text Decides More Than You Do

The length of a fixation tracks the difficulty of what you're processing at that instant, not your general ability: Just and Carpenter showed in 1980 that the eye lingers exactly on rare words and on the spots where a sentence gets complicated. And the single strongest predictor of speed, out of an entire battery of verbal skills tested by Kuperman and Van Dyke (2011), is how fast you recognize words. In other words: on a text full of terms you don't have in your vocabulary, you're measuring your vocabulary, not your reading.

From this, four conditions for the test text.

Unread before. A text you've already been through is recognition, not reading. You'll fly through it and think you've gotten fast.

Known, constant difficulty. Simplest option: two chapters from the same book, same author, same register. One now, one six weeks from now. Fiction and non-fiction differ by around 20 words per minute even at the group-average level (Brysbaert, 2019), and between a novel and a technical manual, the personal difference can top 100.

At least 500 words, ideally 800 to 1,000. A 150-word passage is over, at an average pace, in about 35 seconds. That's how much of a sample a single paragraph gives you: one tangled sentence shifts your number by 20%, and you have no way of knowing how much of the result is just that. Past 500 words, local difficulty starts averaging itself out.

Flowing prose. No bullet lists, no tables, no clipped dialogue. Half-empty lines throw off both the word count and the rhythm, and they inflate your result without you actually reading any faster.

What isn't a test text: an article open on your phone with ads between paragraphs, or anything picked because it's an easy read.

How to Count Words Without Counting Them

Nobody counts 1,200 words with a finger. You estimate, and the estimate is good enough.

Take five full lines from the middle of the page and count the words in each: 12, 10, 13, 11, 11. That's 57 across five lines, so 11.4 words per line. Count the lines on a full page — say, 34. That comes out to roughly 388 words per page.

There's one more correction worth making, because every paragraph ends on a partial line, on average half-empty. Six paragraphs on a page means about three lines to subtract, so your page comes to around 354 words.

The rest is arithmetic. You read three pages plus twelve more lines: 3 × 354 + 12 × 11.4, which comes to about 1,199 words. The stopwatch shows 4 minutes and 22 seconds, or 262 seconds. 1,199 divided by 262, times 60, gives 275 words per minute.

The method carries an error of three to five percent, which doesn't matter: the differences you actually care about run in the tens of percent. What matters instead is using the same method every time — otherwise you're comparing two errors against each other.

Two small details ruin everything if you skip them. Mark, on the spot, in pencil, exactly where you stopped; "somewhere around the middle of the page, I think" adds an error you don't need. And flip your phone screen-down once you start the timer: one running in plain sight turns into an opponent, and you end up chasing the second hand instead of reading.

If you have the text in digital form, the shortcut is obvious: paste the passage you read into a word counter and get the exact figure.

Without a Comprehension Check, You're Measuring Speed, Not Reading

Carver (1990) put it directly: speed without comprehension isn't reading, it's page-flipping. A measurement that only gives you words per minute is, literally, half a measurement.

And the feeling of having understood doesn't substitute for a check. In the report commissioned by NASA Ames, the expert photoreading practitioner rated her own comprehension of the material at 4.5 out of 5; on the actual test she scored 35% on multiple choice, where pure guessing gets you 25% (McNamara, 2000). The feeling that you've understood is the worst measuring instrument you own.

The at-home check takes one minute and works like this. Close the book and, without looking back, state out loud or in writing four things: what the author was arguing, two details that support the idea, and one thing you didn't know before. Only then do you reopen the text and confirm each claim. Rephrase in your own words; a sentence reproduced verbatim only proves it stuck in your ear.

There are three possible verdicts. If at least three out of four hold up, the number is yours. If only two hold up, you read in a rush and the number is inflated. If you can't produce the statements at all, you haven't measured a reading speed — you've measured a page-turning speed.

The honest part: the at-home version doesn't give you a percentage. You're grading yourself, on questions you invent after already knowing the answer, which is a lenient judge by construction. That's why the recall happens before you look back at the text, not after, and why the question is "what was the author arguing," not "did I like it."

Time of Day, Place, and What You Did the Hour Before

There's no published best time of day for reading. What matters is that it's always the same time — preferably whenever you'd read anyway, since that's the rhythm you actually want to track over time. The same logic holds for place: the armchair in the living room and the desk chair don't produce the same number, and you're comparing days, not rooms.

The rest of the conditions are mundane, which is exactly why they get skipped:

• Zero interruptions. A message you glance at for 25 seconds, inside a four-minute measurement, is 10% error — plus the time it takes your attention to get back to the page. There's no way to subtract either one accurately.
• Not right after a pacing drill. You're warmed up, and a second pass through material almost always comes out better, for reasons that have nothing to do with progress.
• The same line width. Perceptual span is fixed, at around 14 characters to the right of the fixation point (Rayner, 1998). A narrow column and a wide one demand a different number of stops per line, at the same underlying ability.
• Three measurements, on three different days. Take the median, not the best one. The best one is, by definition, your lucky day.

What Voids the Measurement

The short list of things after which the number you get isn't comparable to anything:

• text you've already read or skimmed before
• text visibly easier or harder than the one from your previous measurement
• a passage under 500 words
• skipping paragraphs that look hard
• interruptions, however brief
• measured after two hours of reading, when fatigue is already baked into the number
• the stopwatch left running during the comprehension check
• the stopping point estimated from memory
• reading "like it's a test," faster than you normally would

The last one deserves emphasis. If you sprint on the first measurement, you set a benchmark you'll never hit again, and every number after that will look like a regression. A useful starting point is the one you get reading exactly the way you would if nobody were timing you.

What to Do With the Number

Log it with its full context: the date, the time, the book, how many words you covered, how many of the four statements survived the check. Without context, a month from now you're left with an orphaned number.

The next measurement happens four to six weeks out, with the same protocol, on a comparable text. More often than that doesn't make sense: the real gains are small, and the day-to-day noise is bigger than they are, so you'll read progress into the data that isn't there. How much a number has to move to mean something is a discussion of its own, and worth having before you get excited.

And if the result is below what you were hoping for, the useful question isn't "how do I read faster," but which of the brakes is yours: the voice in your head, regressions back along the line, attention that wanders, or vocabulary. The number alone won't tell you which. It only tells you it's worth finding out.

→ Measure your speed and comprehension in a few minutes