Does speed reading work?
You can read faster, but you'll understand less. Going from around 250 words a minute to 500 or 750 while keeping the same grasp of the text is unlikely, according to a 2016 review of the evidence by five reading scientists led by Keith Rayner in Psychological Science in the Public Interest. Speed and accuracy trade against each other. The review's advice for getting through text quicker without losing it is plain: read a lot and build your vocabulary, because language skill is what sets reading speed. (Your own pace is a separate question, covered in reading speed in words per minute.)
Famous fast readers
Evelyn Wood, whose Reading Dynamics course started in 1959, claimed a professor who read more than 2,500 words a minute with excellent recall, and John F. Kennedy was said to manage 1,200. When Slate checked these figures in February 2000, Ronald Carver of the University of Missouri-Kansas City guessed that Kennedy read at 500 to 600 and could perhaps skim at 1,000. Carver called all three claims bunk. The Slate writer, Timothy Noah, added that Wood's professor, if he existed, was probably skimming too.
Carver had tested this himself, Slate reported, in a 1985 study in Reading Research Quarterly. None of his 16 subjects could read faster than 600 words a minute while keeping more than 75% of the content. He also threw out an earlier report of a reader at 17,040 words a minute, since the times, the word counts and the comprehension tests had all been self-reported and self-graded.
Apps that keep your eyes still
Some speed-reading apps flash one word at a time in the same spot, so your eyes never have to hop along the line. The technique is called rapid serial visual presentation, or RSVP. In November 2026 the Spritz website promises “Read 1000+ Words Per Minute” and says retention is better than with ordinary reading, without citing a study.
Eyes don't only move forward, though. Readers jump back to words they've already passed, and those regressions help them understand. Schotter, Tran and Rayner showed this in 2014 by masking each word as soon as the reader's eyes had moved past it, so nothing could be reread, much as in RSVP. Comprehension dropped, and not only on sentences built to be ambiguous. A 2017 study by Acklin and Papesh in the American Journal of Psychology then tested RSVP itself at 700 and 1,000 words a minute against ordinary static text, using passages written at 6th-grade and 12th-grade level. Static text gave the best comprehension. Between the two RSVP speeds the slower one helped verbatim recall, and the faster one did better on questions that needed an inference. Bionic Reading, which bolds the start of each word, didn't speed up skilled adult readers in tests either, although one training study of 60 Iranian learners of English, published in 2025, reported faster reading after eight sessions with the app.
The voice in your head
Speed-reading teachers go after the inner voice. Iris Reading, which sells courses, warns that if you pronounce every word "you will only read as fast as you talk", and its page on reducing subvocalization says that programs which tell you to eliminate the voice completely are wrong, and teaches you to say only some of the words in your head instead. Most readers have one to lose. In a 2017 survey of 570 people by Vilhauer at New York University, 80.7% heard an inner voice at least sometimes when reading silently, and a third of those heard it every time. Most gave it particular qualities, such as a gender, an accent or a pitch, and some heard other people's voices rather than their own. The remaining 19.3% said they understood the words without hearing anything.
Researchers can block inner speech with a technique called articulatory suppression. In a 1990 Oxford study of braille readers it significantly cut comprehension of difficult texts, though speed held up. With easy texts the best readers weren't affected at all and beginners lost both speed and comprehension. Readers in between lost some comprehension and, if anything, read a little faster.
Where skimming pays off
Rayner's review doesn't call fast reading useless. If you don't need a thorough understanding, it says, skimming gets you through a text faster with moderate comprehension.
The loss of comprehension can be small when the questions are easy. Fitzsimmons and colleagues had 32 young adults read 40 Wikipedia articles, half carefully and half by skimming, for a 2020 PLOS ONE study. The skimmers skipped more words and fixated for less time, and still answered 86% of yes/no questions correctly, against 91% when they read with care.
What skimming buys depends on what you're asked afterwards. In 2009 Duggan and Payne at the University of Bath gave readers only enough time to read half of each text. Skimming the whole text beat reading half of it for memory of the important ideas, but it did nothing for the less important details or for inferences, and in a second experiment it was no better than reading the first or second half of every paragraph. The eye tracker in that experiment showed where skimmers go: early in each paragraph, near the top of each page, and to the earlier pages of the document.