Speed reading is one of the few areas where the research is genuinely settled and the popular belief has not caught up. The 2016 review by Keith Rayner and colleagues in Psychological Science in the Public Interest went through the entire literature and reached a conclusion that leaves very little room: there is a trade-off between speed and comprehension, it is not avoidable by training, and the claims made by speed reading programmes are not supported.
That is worth knowing before spending money or hours on it. It is also not the end of the story, because there are ways to get through more material — they are just not the ones being sold.
Why the ceiling exists
Normal adult reading runs roughly 200–300 words per minute with good comprehension. The limit is not your eyes. It is language.
Reading is not a visual scanning task with a processing step bolted on; it is comprehension that happens to be triggered by visual input. Your eyes move in jumps (saccades) and pause (fixations) roughly four times a second, and the pauses are where meaning is built. The clearest evidence for this is that fixation times lengthen on rare words, ambiguous syntax and unfamiliar concepts — the eye slows down exactly where the mind is working harder. You cannot outrun the part that understands.
Two specific claims fail on this basis:
Peripheral vision does not read. The area of the retina with enough acuity to identify words is small — a handful of characters at normal reading distance. Techniques promising you can absorb whole lines or paragraphs at a glance are describing something the eye is not physically equipped to do.
Subvocalisation is not the villain. Suppressing the inner voice is the central promise of most speed reading courses. The inner voice is closely tied to how comprehension works for anything complex, and suppressing it reliably damages understanding of exactly the material you most wanted to understand. It is largely harmless for very easy text, which is why the demonstration always uses very easy text.
RSVP does not solve it either. Apps that flash words one at a time in a fixed spot remove eye movement, which sounds like it should help. It removes something important instead: the ability to look back. Regressions — flicking back to a phrase you did not resolve — are not a bad habit, they are a repair mechanism, and comprehension of anything demanding drops without them.
What the honest numbers look like
Around 400–500 words per minute appears to be the practical upper bound for reading with real comprehension, and getting there is mostly about vocabulary and familiarity rather than technique. Claims of 1,000+ words per minute with full understanding describe skimming, which is a legitimate skill with a different name and different results.
The reliable predictor of reading speed, unglamorously, is how much you already know about the subject. An expert reads a paper in their field fast because they can predict most of it. That is not a technique you can install in a weekend; it is the accumulated effect of having read a lot.
What actually helps
Skim deliberately, and know you are doing it. Skimming is genuinely useful and the research does not dispute that — you get the gist, you miss detail, and you should choose it consciously rather than believe you are reading. The three-pass structure in how to read a research paper is skimming with a decision attached.
Read less, not faster. The largest available speed gain is not reading things you did not need. Most reading lists are aspirational rather than necessary, and abandoning a book at chapter two is a skill worth developing. See the antilibrary and read it later.
Preview before reading. Two minutes on the structure — headings, first sentences, conclusion — measurably improves both speed and comprehension on the real pass, because you are no longer building the map while walking. This is most of what SQ3R and PQ4R are doing.
Build vocabulary and background. The most effective long-term intervention, and the least marketable.
Match speed to purpose. Entertainment, gist, working knowledge and mastery are four different targets with four different appropriate speeds. Applying one speed to all reading is the actual inefficiency.
The part everyone skips
Reading faster is the wrong goal for most people anyway, because the bottleneck is not intake. It is that almost none of what you read survives the week.
You will forget the large majority of a book within days without any retrieval — this is the forgetting curve, and it applies at 200 words per minute exactly as it does at 600. Doubling your reading speed doubles your rate of forgetting, which is why heavy readers frequently cannot say much about books they finished last year.
What changes retention is what happens after: closing the book and reconstructing the argument, writing the claim in your own words, and revisiting it once or twice. That is active recall and the generation effect, and ten minutes of it beats an hour of faster reading. See how to remember what you read and how to write a summary.
The practical version: read one book properly and write four sentences about it, rather than skimming five and retaining nothing. The second option feels far more productive and produces almost nothing you can use.
The short answer
You cannot train yourself past the comprehension ceiling, and the courses selling that are selling skimming under another name. You can read the right things, preview before you start, skim on purpose, abandon more books, and process what you read so it survives.
That is slower on paper and considerably faster in terms of what you actually end up knowing.
More in how to take notes from a book, how to highlight effectively, and memory and focus.