How to Remember What You Read (What Actually Works)

Forgetting most of a book within days is normal. That is the classic forgetting curve, mapped by Ebbinghaus over a century ago: memory for new material fades fast at first, then levels off. The fix is not reading slower — the best-evidenced way to remember what you read is retrieval practice, meaning you close the book and recall what it said, which beats rereading for long-term retention (Roediger and Karpicke, 2006). Spaced review is a close second.

That is the whole answer, compressed. The rest of this article is those two techniques plus three more, ranked by how much evidence backs each one — and an honest look at why the two most popular study habits on the planet barely work at all.

Why do you forget what you read?

Because your brain treats reading as recognition, not storage. While you read, every sentence feels understood — and it is, in the moment. But understanding a sentence as it passes and being able to produce its content three days later are different operations, and the second one never happens automatically. The forgetting curve is not a design flaw. It is your memory doing triage on an enormous stream of input, and by default a book you read once, passively, does not make the cut.

The cruel part is that reading feels like learning. Fluent prose produces a sense of mastery that has nothing to do with what you will retain. It is the same gap we keep pointing at in the average reading speed piece: the feeling of reading well and the measurement of reading well are different instruments, and the feeling flatters you.

Here is the arithmetic that should change your behavior. A 90,000-word non-fiction book, read at the adult average of 238 WPM (Brysbaert, 2019), takes about 378 minutes — a bit over six hours. (The division is ours; check it.) If you do nothing after those six hours, the curve does its work and most of the detail is gone within days. Five minutes of active recall after finishing is under 2% added on top of the time you already invested — and it is the part that decides what you keep. Skipping it is like cooking for six hours and not eating.

What actually works? Five techniques, ranked by evidence

Ranked from strongest evidence to weakest. The first two have decades of research behind them. The last three are well-grounded but softer — they work to the degree they smuggle in the first two.

1. Test yourself (retrieval practice)

The single best-supported technique in the literature. In Roediger and Karpicke's 2006 study in Psychological Science, people who recalled a passage from memory retained it better on delayed tests than people who reread it instead. Pulling information out of memory strengthens it in a way that pushing it back in does not — even though rereading is the option that feels more productive, which is exactly the trap.

The practical version costs almost nothing. Finish a chapter, close the book, and write down everything you remember on a blank page — no peeking. Then open the book and check. The recall attempt itself is the treatment; the check just keeps you honest. Three to five minutes per chapter.

The blank page will humiliate you at first. That is the technique working. The gaps you find are exactly the parts that were about to slide off the curve.

2. Space your reviews

The second pillar. Cepeda and colleagues' 2006 meta-analysis in Psychological Bulletin found that spreading study sessions over time produces better retention than massing the same time into one sitting. Three hours split across three evenings beats three hours in one sitting — same total effort, different memory.

A practical default schedule, which is our suggestion rather than anything prescribed in the study: recall the material the day after you finish, again about a week later, again about a month later. Each review is the blank-page exercise from technique one, not a rereading session. The gaps should grow. Precision matters less than the existence of gaps — a sloppy spaced schedule beats a perfect crammed one.

3. Take notes in your own words

Notes help in proportion to how much rewriting they force. Copying a sentence out of the book is transcription — your hand works, your memory does not. Rephrasing the idea in your own words forces you to process the meaning rather than the wording, and it doubles as a small act of retrieval, since you have to hold the idea in mind while you reconstruct it. That is the honest reason it works: it borrows its power from technique one.

A good filter: if your note could have been produced by a photocopier, it is not doing anything.

4. Connect it to what you already know

Memory is associative. An idea linked to things you already understand has many retrieval routes; an isolated fact has one, and the curve takes it first. So interrogate the material as you go: what does this confirm, contradict, or complicate about what I already believed? Where have I seen this pattern before? An idea that picks a fight with your existing beliefs is an idea you will still have next month.

5. Explain it to someone

Teaching is retrieval with the stakes turned up. Explaining a chapter to a friend forces you to recall it, organize it, and simplify it, all at once — and every hand-wave becomes audible. No willing friend is required: writing the explanation as if for one works nearly as well, because the gaps become just as obvious. This ranks last only because it is the most expensive, not because it is weak.

Baseline first — measure your speed and comprehension free →

Why don't rereading and highlighting work?

Now the uncomfortable part. Dunlosky and colleagues reviewed the major study techniques in a 2013 monograph in Psychological Science in the Public Interest and rated each by utility. Practice testing and distributed (spaced) practice came out on top. Rereading and highlighting — comfortably the two most popular study habits in existence — rated low utility.

Why do the losers stay popular? Because they are easy and they feel productive. Rereading produces fluency — the text flows more smoothly the second time — and your brain misreads that smoothness as knowledge. Highlighting feels like an act of selection, but dragging a marker across a sentence requires no processing of what the sentence means. Both let you touch the material without ever asking your memory to produce it. They are the study equivalents of reorganizing your desk.

If you are a student staring down actual exams, this finding matters even more, and we have turned it into a study-specific playbook in speed reading for students.

Should you read slower to remember more?

Not by itself, and this surprises people. Slowness is not a memory technique. A slow, passive pass through a chapter is just an extended act of rereading-as-you-go, and we have already seen where rereading ranks. The lever is not pace — it is whether retrieval and spacing happen at all.

The reverse direction does have a limit, though. Rayner and colleagues' 2016 review found that comprehension declines as reading speed is pushed, with the practical ceiling for real reading somewhere around 400 to 500 WPM. Past whatever pace your comprehension holds at, you are not reading faster — you are remembering less per minute. You cannot retain what you never comprehended in the first place; retrieval practice on a chapter you skimmed retrieves fog.

So the honest formula is this: read at the fastest pace where your comprehension stays intact — for most adults that is somewhere between the 238 WPM average and the low 400s — and then spend a slice of the time you saved on recall. On the evidence above, bet on the reader who finishes a chapter at 300 WPM and does three minutes of blank-page recall over the one who crawled through it at 150 WPM and did nothing. Speed buys you time; retrieval decides what the time was worth.

This is also why a speed number without a comprehension score is vanity. Our free reading speed test measures both, because either number alone is close to meaningless — a point we beat to death in reading comprehension and speed.

Get both numbers — speed and comprehension, 2 minutes →

A system that survives real life

Everything above, compressed into a protocol small enough to actually happen:

1. At every chapter end: look away and say or scribble three things you remember. Thirty seconds. This is retrieval practice at its cheapest.
2. Same day you finish: one blank-page summary in your own words, under ten minutes. No peeking until you are done.
3. On the calendar: two more recalls — one about a week out, one about a month out. Recall first, check the book second.
4. Make it automatic: decide in advance, in the form of an if-then plan. Gollwitzer's work on implementation intentions found that plans shaped as "when situation X arises, I will do Y" increase follow-through compared with vague intentions. So: "When I finish a chapter, I will close the book and write three things I remember." Specific trigger, specific act.

If this looks like a stripped-down study system, it is. SQ3R — the grandfather of them all, introduced by Robinson in 1946 — endures for exactly one reason: its Recite and Review steps are retrieval practice and spaced review wearing 1940s clothes. You do not need the whole apparatus. You need the two steps that were carrying it.

One more thing this unlocks: volume finally pays. Reading more books only compounds if you keep something from each one, which is why this article and how to read more books are really two halves of the same argument.

The takeaway

You will still forget most of what you read. The curve always wins on volume, and any advice claiming otherwise is selling something. What you can control is which parts survive: the ideas you retrieved, spaced, rephrased, connected, and explained will still be there in six months, and the rest was rental.

That reframe matters more than any single technique. Stop measuring yourself by how much slipped away — that number is rigged against you by design. Measure the five minutes of recall you did or did not do after closing the book. It is the smallest input in your whole reading life, and it is the one that decides what you get to keep.