A words-per-minute score tells you how quickly you got to the end of a text. It does not by itself show how much of the text you took in. Research finds a trade-off between speed and comprehension: pushing substantially beyond a comfortable reading pace tends to reduce understanding (Rayner et al., 2016). That is why a speed figure is more useful alongside a comprehension check from the same reading.

The trade-off is a property of reading, not a personal failing

Rayner et al. (2016) reviewed decades of research on how people read and on the claims made for speed reading. Their central conclusion is a speed–accuracy trade-off. Readers can move faster through a text, but they do it by processing less of it, and their understanding falls accordingly. The review found no good evidence that people can double or triple their speed while keeping comprehension at the level of normal reading.

This matters for how you interpret any number. Suppose two people read the same 400-word passage. One finishes in 1 min 20 sec, which is 300 WPM. The other takes 1 min 49 sec, about 220 WPM. If the first reader can answer half the questions about the passage and the second can answer all of them, calling the first one “the faster reader” misses what happened. They did different amounts of work.

Slowing down is the processing

It is tempting to treat the time your eyes spend on a word as dead time that training could remove. The eye-movement research points the other way.

Just & Carpenter (1980) recorded where readers looked while reading passages and found that the time spent on each word tracked the mental work the word required. Readers lingered on infrequent words and at points where the pieces of a sentence had to be combined into a meaning, including the ends of sentences. Their account treats the pause as the visible side of comprehension: the eyes stay on a word while the mind is still dealing with it.

Rayner (1998), reviewing twenty years of this research, describes the same pattern from the other direction. When text becomes harder, fixations get longer, the jumps between them get shorter, and the eyes go back more often. A reader does not choose this. It is how a skilled reading system responds to difficulty.

So when you slow down on a hard paragraph, you are not reading badly. You are doing the part of reading that produces understanding.

Going back to re-read supports understanding

Speed-reading advice often lists “regressions”, the backward eye movements to earlier words, as a bad habit to eliminate. Rayner (1998) reports that they are a routine part of skilled reading, making up roughly 10–15% of eye movements.

Schotter, Tran & Rayner (2014) tested what they are for. In their experiment, words were masked once the reader’s eyes had moved past them, so looking back returned nothing useful. Comprehension dropped compared with normal reading, where looking back was possible. The implication is direct: readers use regressions to repair a misreading or check something they were unsure of, and when that repair is taken away, they understand less.

This is also the main objection to apps that flash one word at a time. That presentation removes the option to look back. Rayner et al. (2016) conclude that it tends to reduce comprehension of longer texts for that reason, among others.

What skimming keeps and what it loses

None of this means that reading faster is always wrong. Skimming is a real and useful skill; it just produces a different result from reading.

Duggan & Payne (2009) studied people who had to get through text under time pressure. Their findings suggest that skimmers who sample a text tend to pick up the more important ideas from the parts they look at, which is a sensible return for limited time. The cost is detail. Material that was skipped, and the finer points of what was sampled, are less likely to be retained. Rayner et al. (2016) make the same point: skimming is a reasonable choice when the gist is enough, and a poor one when the details matter.

The practical question is therefore not “how fast can I go?” but “what do I need from this text?” A contract clause, a set of instructions and a chapter you will be examined on all need the details. A newsletter you are checking for one relevant item does not.

How to read the two scores together

The site’s reading speed test reports both numbers for this reason. You choose one of five original passages of roughly 300–450 words. The timer starts when you click Start Reading and stops when you click Finished Reading, and then you answer four multiple-choice questions about what you read. The result shows your raw WPM, your time, your comprehension quiz score, and a comparison with the adult English reading averages reported by Brysbaert (2019): 260 WPM for the narrative passage and 238 WPM for the other passages.

A rough guide to the combinations:

Speed Quiz result How to use the result
Moderate 4 of 4 An encouraging attempt; compare it with other passages
High 4 of 4 Try a different passage before treating this as a usual pace
High 2 of 4 or fewer Do not count speed alone as progress; try a slower attempt
Low 2 of 4 or fewer Difficulty, distraction or a misunderstood question may matter; retest

The third row is a reason for caution. High WPM with a weak quiz result does not establish useful reading speed. Equally, four answers cannot tell us exactly what caused the result or what fraction of the whole passage was understood.

The comprehension-adjusted score, with a worked example

The test also offers an optional comprehension-adjusted score. The calculation is simple: raw WPM multiplied by the fraction of questions answered correctly.

Take a reader who finishes at 300 WPM and gets three of four questions right. Three of four is 0.75, so the adjusted score is 300 × 0.75 = 225.

Compare a second attempt by the same reader at a calmer pace: 240 WPM with all four correct. The adjusted score is 240 × 1.00 = 240. The slower attempt scores higher because a larger share of the quiz questions was answered correctly. A third attempt at 340 WPM with two of four correct gives 340 × 0.50 = 170. Its adjusted figure is lower, although these attempts alone do not establish that pace caused the score difference.

That is what the score is for: it stops a raw speed gain from looking like progress when comprehension fell by more than speed rose.

Limits of the adjusted score

The adjusted score is a convenience metric of this site. It is not a standardised measure and it is not used in the research cited here. Treat it as a quick sanity check, with these limits in mind:

  • Four questions is a coarse measure. Your comprehension score can only be 0, 25, 50, 75 or 100%. One misread question moves the adjusted score by a quarter of your raw WPM.
  • Multiple-choice questions can be guessed. A correct answer does not always mean the point was understood.
  • Questions sample the passage. They cover a few points, not everything. You can miss a detail that was not asked about, or be asked about the one sentence you skimmed.
  • Prior knowledge helps. If you already know the topic, you may answer correctly without having read closely.
  • One passage is one sample. A single short text does not represent everything you read.

Because of this, look at patterns across several attempts and passages rather than any single result. Our guide on how to measure reading speed covers how to set up a fair test, and the guide to what WPM means in reading explains the raw figure itself.

What to do with this

Decide what you need from a text before deciding how fast to go. For anything where details matter, your useful speed is the fastest pace at which you still understand it, and that pace is slower for hard material than for easy material.

When you test yourself, record both numbers every time. If you are trying to judge whether your pace is reasonable, read our guide on what counts as a good reading speed, which makes the same point from the other side: a good speed is one that fits the text and the purpose. And if you are wondering whether training can remove the trade-off altogether, the evidence is summarised in does speed reading really work.