Does speed reading work? What the research actually says
Speed reading has a sixty-year track record of big promises and a thinner one of measurements. Here is what controlled studies found, technique by technique, and what an honest app can promise.
By Mario Diaz, founder of Black Box Code LLC. .
In short
Speed reading works only in a narrow sense. The canonical 2016 review by Rayner, Schotter, Masson, Potter and Treiman concluded that speed-reading claims are overstated: there is an unavoidable trade-off between speed and accuracy, and no documented reader reaches 1,000 words per minute with full comprehension. Techniques that suppress subvocalization, expand peripheral vision or bold parts of words show no benefit in controlled tests, and some hurt comprehension. What does work is modest: paced reading slightly above your measured speed with comprehension questions gating progression, timed reading with quizzes, and repeated reading of the same passage. Those methods produce gains of roughly 40 to 110 words per minute over a term, moving a typical reader from about 240 WPM to 250 to 400 WPM with comprehension held or improved. That is a real 30 to 70% gain. Anything past 500 to 600 WPM is skimming.
The review that settled the big claims
In 2016 Psychological Science in the Public Interest published "So Much to Read, So Little Time: How Do We Read, and Can Speed Reading Help?" by Keith Rayner, Elizabeth Schotter, Michael Masson, Mary Potter and Rebecca Treiman. It is the most thorough review of the field and its conclusions are blunt: speed-reading claims are "overstated", there is an unavoidable speed-accuracy trade-off, and the honest route to faster reading is growing skill and vocabulary through practice.
The constraint is physiological. The eyes fixate for about 200 to 250 milliseconds, jump in 25 to 30 milliseconds, and identify roughly one word per fixation, with a perceptual span of about 14 characters to the right of the fixation point (McConkie and Rayner, 1975). Multiply that out and word-by-word reading tops out in the 500s.
Technique by technique
Subvocalization suppression: debunked
Baddeley, Eldridge and Lewis (1981) had readers suppress inner speech while reading. Comprehension got worse and speed did not improve. Inner speech is part of parsing sentences, not a bottleneck to remove.
Peripheral vision expansion: impossible
Courses that teach "read a whole line at a glance" ask the retina to identify text it cannot resolve. The perceptual span allows two or three words, which is why phrase-level chunking of two to three words is defensible and anything larger is not.
Bionic-style bolding: no effect
Bolding the first letters of words was tested by Snell (2024, Acta Psychologica) and in a 2022 field test by Readwise with 2,074 readers. Neither found a reading-speed benefit.
RSVP (one word at a time): useful in a narrow band
Rapid serial visual presentation removes eye movements by flashing words at a fixed point. Benedetto et al. (2015) found it raised visual fatigue and hurt literal comprehension compared to a page of text; Acklin and Papesh (2017) found comprehension collapsed at 700 to 1,000 WPM. Used as a pacing drill between 100 and about 450 WPM, with a way to rewind, it has a place. More on RSVP reading.
Blocking regressions: harmful
Going back to re-read is often sold as a bad habit. Schotter, Tran and Rayner (2014) prevented readers from re-reading and comprehension fell from about 75% to chance. Regressions are repair work. Reduce the unnecessary ones with a pacer; never block them.
What does work
- Adaptive paced reading. Breznitz's Reading Acceleration Program advanced text slightly faster than each reader's own rate, with comprehension questions gating every step up. Readers got faster, made fewer errors, understood more, and kept the gains at a six-month follow-up (Breznitz, 2013; Horowitz-Kraus, 2014).
- Timed reading with quizzes. Chang and Millett (2013) measured gains of 40 to 50 WPM per term; Cushing-Weigle and Jensen (1996) measured 110 WPM over ten weeks with no comprehension loss.
- Repeated reading. Therrien's 2004 meta-analysis found large fluency and comprehension effects that transferred to new passages.
- Comprehension-gated progression, everywhere. Without questions, readers drift into skimming while believing they still understand (Rayner et al., 2016).
What an honest app should promise
From a 200 to 250 WPM baseline: 250 to 400 WPM on easy-to-moderate text with comprehension at or above baseline. That is a 30 to 70% gain, achieved over weeks, and it depends on the reader and the text. Skimming past 500 WPM is a separate skill worth learning for the right material, and it should be called skimming.
ReadSwift was designed from this list. It builds the four things that work, leaves out the four that do not, shows effective WPM (speed multiplied by comprehension) as the only headline number, and marks 500+ WPM as the skimming zone on every screen.
Questions about this
Can anyone read 1,000 words per minute?
Not with full comprehension. No documented case exists in the research literature. Readers tested at those rates were skimming: they could report the gist but not the details, and did no better than untrained people told to skim.
Is speed reading a scam?
The big claims are. The modest version, paced practice slightly above your speed with comprehension checks, is well supported and produces a real 30 to 70% gain over weeks. The difference is whether the method measures comprehension or only speed.
Which speed reading technique is backed by science?
Adaptive paced reading with comprehension gating (Breznitz), timed reading with quizzes (Chang and Millett; Cushing-Weigle and Jensen) and repeated reading (Therrien). Subvocalization suppression, peripheral-vision expansion and bionic bolding are not.