Speed reading works in a narrow sense and fails in the sense that is usually advertised. Training can make you faster at skimming, which is useful when the gist is enough. The claim that you can double or triple your reading speed and still understand the text as well as before is not supported by the research (Rayner et al., 2016).

A major review of the question is by Rayner and colleagues, a group of reading scientists who examined both how reading works and what speed-reading methods claim. Most of what follows is based on it.

What speed-reading courses and apps claim

The methods differ, but the core claims are similar:

  • Ordinary readers are slow because of bad habits, not because of any real limit.
  • The inner voice that “says” words as you read holds you to speaking pace and can be switched off.
  • Looking back at earlier words wastes time and can be trained away.
  • With practice you can take in several words, a whole line or a block of text at each glance.
  • Technology can remove eye movements altogether by flashing words one at a time.

Each claim points at something real. Readers do subvocalise, eyes do go back, and eye movements do take time. The disagreement is about whether these things are waste or part of how understanding is produced.

What the visual system allows

The first constraint is the eye. Sharp vision is limited to a small area at the centre of the visual field, and clarity falls off quickly on either side. Rayner (1998), reviewing eye-movement research, reports that for readers of English the region from which useful information is taken in at each fixation, the perceptual span, reaches only about 3–4 letters to the left of the fixated point and about 14–15 letters to the right. Words are identified within a smaller part of that region.

This is why readers move their eyes in small jumps and fixate most of the words. It is also why “reading a whole line at once” or “reading down the middle of the page” is not reading in the usual sense. Words that never fall near the centre of vision are not identified. Rayner et al. (2016) found no good evidence that training enlarges the span enough to change this.

The larger limit is language, not eyesight

Even if the eyes could deliver words faster, they would still have to be understood. Rayner et al. (2016) conclude that the main bottleneck in reading is the ability to recognise words and build meaning from sentences, not the mechanics of eye movement. Faster readers tend to be people with stronger language skills and larger vocabularies, not people with unusual eye movements.

This reframes the whole question. If the limit were the eyes, eye training would help. Since the limit is mostly comprehension, moving the eyes faster just delivers words more quickly than they can be processed.

Subvocalisation

Most people experience a quiet inner voice while reading. Speed-reading methods often treat it as the main obstacle. The research summarised by Rayner et al. (2016) indicates that this inner speech plays a part in comprehension, helping readers identify words and hold a sentence in mind while its meaning is worked out. When experiments interfere with it, understanding tends to suffer, especially for harder text.

The inner voice is also not a literal reading-aloud. Adults reading silently average well above speaking pace: Brysbaert (2019) puts silent reading of English non-fiction at 238 WPM and reading aloud at 183 WPM. So subvocalisation does not pin silent reading to speech rate in the way the claim suggests.

Regressions

Backward eye movements look like inefficiency. Schotter, Tran & Rayner (2014) tested this directly by masking words once a reader’s eyes had moved past them, which made looking back pointless. Comprehension was worse than in normal reading. Readers go back because they have misread something or need to check it, and without that option the error stays in place.

A reader trained never to regress has not become more efficient. They have agreed to carry on without fixing mistakes.

RSVP apps

Rapid serial visual presentation shows one word at a time in the same spot, at a rate you set. Because the eyes do not need to move, the rate can be pushed very high.

Rayner et al. (2016) raise several problems. The reader can no longer look back. Every word gets the same time, although hard words need more than easy ones. And the brief preview of the next word that normal reading provides is gone. Their conclusion is that these apps tend to reduce comprehension of longer texts. For a short notification or a single sentence the cost may be small. For a chapter it is not.

What about very high claimed speeds

Claims of reading at several times the normal rate with full understanding have been tested. The review by Rayner et al. (2016) concludes that when such readers are examined carefully, their comprehension is not at the level of normal reading: they do reasonably on the general gist, particularly of familiar material, and poorly on details and on unfamiliar or difficult text. In other words, what is being demonstrated is skilled skimming, often supported by prior knowledge of the subject.

The arithmetic shows why the claims attract people. An 80,000-word book takes 5 hr 36 min at the 238 WPM non-fiction average reported by Brysbaert (2019). Tripling that pace to 714 WPM would bring it down to about 1 hr 52 min. The time is real. The question is what is left of the book afterwards, and the evidence says it is the outline, not the content. For context on ordinary rates, see our guide to average reading speed.

Where proponents have a point

A fair summary should include what speed-reading advice gets right.

  • Skimming is a legitimate skill. Duggan & Payne (2009) studied people skimming under time pressure, and their results suggest that skimmers tend to pick up the more important ideas from the parts of the text they sample. Detail is lost, but for many purposes that is an acceptable exchange. Our guide to reading reports, emails and work documents shows how to decide what to skim and what to read in full.
  • Many people read everything at one pace. Being encouraged to go faster on easy or low-value material is reasonable advice, even if the explanation given for it is wrong.
  • Previewing helps. Looking at headings and summaries before reading is common in speed-reading courses and is sensible practice.
  • Practice matters. Courses make people read a lot, and reading a lot is what Rayner et al. (2016) identify as the best-supported way to become a faster reader, through better vocabulary and language skill.

The problem is not that these methods do nothing. It is that their effect is described as faster reading when it is mostly a switch from reading to skimming.

Summary of claims and evidence

Claim What the evidence indicates
The inner voice slows you down and should be suppressed Inner speech supports comprehension; suppressing it tends to hurt understanding
Re-reading is a habit to remove Regressions repair misreadings; blocking them lowers comprehension
You can train a wider visual span Only a small window of text is taken in per fixation; no good evidence that training changes this
Flashing words removes wasted eye movement Comprehension of longer texts tends to fall
Speed can be multiplied with no loss Speed and comprehension trade off

What a reader should conclude

Three practical points follow.

First, treat skimming and reading as different tools. Skim when the gist will do, and read at your normal pace when details matter. Neither is better in general.

Second, if you want to read faster with the same understanding, the route supported by evidence is slower and plainer than a course: read more, and build vocabulary and knowledge in the subjects you care about. Our guide on how to improve reading speed sets this out.

Third, check any technique against comprehension rather than taking the speed gain at face value. You can start with the reading speed test on this site. Read one passage normally, then try a second passage with the technique. The test times your reading, asks four comprehension questions and reports raw WPM alongside your score. Higher WPM with a lower quiz score is a reason for caution, but different passages and just four questions cannot establish that the technique caused a change. Compare several unfamiliar passages before drawing even a tentative conclusion. The reasoning is covered in more depth in reading speed versus comprehension.