Study strategy · drilling vs rereading · the adherence gap

Drill questions vs rereading notes: the science is settled, the habit is not

Matthew Diakonov, Written with AI

Published May 19, 20267 min read

You have read the answer to this before. Practice questions beat rereading, it is one of the most replicated findings in learning research, and every study guide ends on the same instruction: stop rereading, make practice questions. And yet here you are, notes open, highlighter in hand, doing the thing the research told you not to do. This page is not another retelling of the science. It is about the gap between knowing drilling wins and actually doing it every day.

Why you keep rereading anyway →

Direct answer · verified May 19, 2026

Drill the questions. In the study most often cited on this, Karpicke and Roediger (2008, Science), students who repeatedly retrieved material recalled about 80% of it one week later; students who repeatedly reread the same material recalled about 36%. Same total time, more than double the retention. The honest catch is not in the science, it is in the doing: rereading only loses if you actually replace it with drilling and keep that up. A method you abandon in week two scores zero, no matter what the lab found.

Research summary and primary-source citations: retrievalpractice.org, maintained by Pooja Agarwal and colleagues.

0%One-week recall, repeated retrieval

0%One-week recall, repeated rereading

0Studyly question-quality eval score

0Turbolearn on the same eval

Retention pair: Karpicke and Roediger 2008 paradigm (40 word pairs, one-week delayed test). Question-quality pair: held-out three-document eval, methodology on studyly.io/quality.

Why this was never really a debate

Rereading is not a dumb choice. It feels productive, and that feeling is exactly the trap. Cognitive psychologists call it the illusion of fluency: on the second pass the words look familiar, your eyes move faster, and your brain reads that familiarity as mastery. You close the notes feeling like you know the chapter.

The exam does not ask you to recognize a page. It asks you to retrieve a fact that is not in front of you. Recognition and retrieval are different skills, and you can only build the second one by trying to do it. That is the whole case for drilling: a question with the answer hidden rehearses the exact move the exam will demand. A reread rehearses the move the exam never asks for.

One rereading rep vs one drilling rep, compared honestly

The two methods do not differ on one axis, they differ on several. Rereading genuinely wins two rows. They are the two rows that explain why people choose it and why it does not work.

Feature	Rereading your notes	Drilling practice questions
Recall one week later (Karpicke and Roediger 2008 paradigm)	About 36% with repeated rereading	About 80% with repeated retrieval
Which skill the rep trains	Recognition. The page is in front of you, so you practice noticing a fact, not producing it.	Retrieval. The answer is hidden, so you practice the exact move the exam scores: pulling a fact from cold memory.
What it tells you about your gaps	Nothing reliable. Rereading feels uniformly good across material you know and material you do not.	Every miss is a labeled gap. You learn which slide you do not know, not just that the chapter feels familiar.
Cost to start a session	Zero. The notes already exist; you just open them.	The questions must exist first. Hand-authoring them is an hour or two per deck, which is the real reason people reread instead.
Cost on a low-willpower night	Low. Comfortable, which is why it survives bad days and does nothing.	Higher. Uncomfortable by design, which is what makes it work and also what makes it easy to skip.
Does it hold up on the fifth pass of the same material	Yes, in the sense that you keep rereading and keep learning nothing new.	Yes, when the stem is auto-rephrased on revisit so you retrieve the fact instead of memorizing the wording.

The first read of new material is rereading's legitimate job. This table is about every pass after that.

The gap nobody writes about: why you keep rereading

Here is the part the guides skip. If drilling is so clearly better, why does almost everyone keep rereading? Not because they have not heard the science. Because rereading wins the only contest that decides which method you actually use: the cost of starting.

Rereading has zero setup. You open the notes that already exist and your eyes start moving. Drilling has a setup tax: the questions have to exist first. Hand-writing 100 questions from a 90-slide lecture deck is an hour or two of work before a single rep happens. So the tired student at 11 p.m. does the arithmetic without noticing they are doing it, and rereading wins because it is the thing you can start in four seconds.

The second cost is willpower. Rereading is comfortable; drilling is uncomfortable by design, because failing to retrieve a fact is the unpleasant moment that actually teaches you. Comfortable habits survive bad days. Uncomfortable ones get skipped, then skipped again, and a method skipped for a week is a method you have quit. This is why most students who switch to a flashcard or recall app are back to rereading by week two. The science never had an adherence problem. The student does.

Two semesters, same student, same lectures

The difference between these two is not intelligence and not effort. It is which habit had a low enough starting cost to survive a whole semester of tired weeknights.

The same lectures under two study habits

Week 1. Lecture ends, you read the slides over coffee, highlight the dense parts. It feels like studying. It is comfortable and it requires nothing to exist first. Week 4. Same routine. The decks are familiar now, your eyes glide. You feel on top of the material because every reread feels easy. That ease is the illusion of fluency, not mastery. Exam week. You sit down and the question hides the slide. The familiarity you built does not transfer to producing the answer. You knew it when you could see it. You blank when you have to retrieve it, and you cannot tell which decks you actually never learned.

Zero setup cost, so the habit never breaks
Builds recognition, which the exam does not test
Familiarity feels like mastery until the slide is hidden

Anchor · what closes the adherence gap

One tree per deck, drilled in about 0 minutes a night

The reason rereading wins is that it costs nothing to start and the work is invisible, so there is nothing to quit. Studyly attacks both. The 60-second conversion removes the setup tax: you upload a deck instead of hand-writing questions for an hour, so the drilling habit is as cheap to start as opening your notes.

Then it makes the work visible. Each lecture deck grows its own tree on your dashboard. Decks chain into a river. A weekly league puts your streak next to other students. None of that is a cognitive-science variable, and that is the point: the science was never the problem. Adherence was. Visible, accumulating progress is one of the most reliable things known for keeping someone from quitting a habit, and it is the mechanic most retrieval apps leave out, which is why they lose users in the first couple of weeks.

That is the checkable claim on this page. Not that drilling beats rereading, which the research settled long ago, but that a tree per deck and a weekly league are the specific features that turn the settled science into a habit you still have in week ten.

81.3

“Generated 200 questions from a 90-slide cardiology deck in 47 seconds. Studyly scored 81.3 on the same eval where Turbolearn scored 57.8.”

Held-out three-document eval, Jungle internal admin Quality Comparison panel, 2026-04-24

How to actually make the swap

You do not have to white-knuckle this. The swap works when each step costs less than the rereading it replaces. Four steps, none of them a marathon.

Read the deck once, then close it

The first read is real and necessary. You cannot retrieve a fact you have never encoded. But one pass is the budget. The moment you finish reading, the rereading is done and the drilling starts.

Convert the deck instead of rereading it

Upload the lecture slides, the PDF, the textbook chapter, or the YouTube lecture. Around 200 questions come back in about 60 seconds, across MCQ, free-response, case-style, and image-occlusion formats. This is the step that removes the hour-or-two setup tax, which is the step that decides whether the habit survives.

Drill five minutes the night of the lecture

Not a marathon. Five minutes, same-day, while the reading is fresh. Same-day drilling is your first genuine retrieval rep. The deck grows a tree, so the five minutes leaves a visible mark instead of vanishing.

Re-drill misses under rephrasing, never by rereading the slide

When a missed question returns, the stem is reworded and the options reshuffled, so the fifth pass is still a retrieval and not a memorized string. On a wrong answer, read only the cited source slide, then close it. Do not reopen the whole deck. That is the rereading habit trying to come back.

When rereading still earns its place

This page is not anti-reading. There is exactly one job rereading does that drilling cannot: the first encoding. You cannot retrieve a fact you have never seen, so the first pass through a new lecture is legitimate, necessary work. The mistake is not the first read. It is the second, third, and fourth read standing in for practice.

One more honest case: if your material is computational, worked equations, dose calculations, organic mechanisms, then a concept-recall question is the wrong tool and you want a step-by-step solver instead. Drilling questions test whether you can name an enzyme; they do not teach you to balance a redox reaction. For everything memorization-heavy, though, the rule holds: read once, then drill.

Drop one lecture deck and skip the hour of writing questions

Upload lecture slides, a PDF, a textbook chapter, or a YouTube lecture. About 200 questions come back in roughly 60 seconds, generated against your source, across MCQ, free-response, case-style, and image-occlusion formats. Free tier on app.jungleai.com, no credit card.

Convert a deck free →

Drilling questions vs rereading notes: common questions

Is rereading my notes ever as effective as drilling practice questions?

For building durable recall, no. In the Karpicke and Roediger 2008 paradigm cited across the learning-strategy literature, repeated retrieval produced about 80% recall one week later versus about 36% for repeated rereading, on the same total study time. The one place rereading is not just acceptable but required is the very first pass through new material: you cannot retrieve a fact you have never encoded. Read each deck once, then everything after that first pass should be drilling, not rereading.

If drilling questions is clearly better, why do I keep rereading?

Because rereading wins the contest that actually decides which method you use day to day: the cost of starting. Rereading has zero setup, you just open notes that already exist, and it is comfortable on a low-willpower night. Drilling needs questions to exist first, which is an hour or two of hand-authoring per deck, and it is uncomfortable by design. Knowing the science does not change behavior; lowering the setup cost and making the habit stick does.

What is the illusion of fluency?

It is the reason rereading feels productive while doing little. On a second pass through a deck the words look familiar, your eyes move faster, and your brain reads that familiarity as mastery. But the exam hides the page and asks you to produce the fact. Recognizing a fact you can see and retrieving one you cannot are different skills. Rereading drills the first; only drilling questions drills the second, which is the one the exam scores.

Where do the practice questions come from if I stop rereading?

That is the real bottleneck, and it is why most students never actually make the switch. Hand-writing 100 questions from a 90-slide lecture deck takes an hour or two, which no tired student will do, so they reread instead. Studyly removes that step: drop in the lecture slides, a PDF, a textbook chapter, or a YouTube lecture and you get around 200 questions in about 60 seconds, generated against that exact source rather than pulled from a generic web question bank.

How does Studyly keep me drilling daily instead of quitting in week two?

By making five minutes a night leave a visible mark. Each lecture deck grows its own tree, decks chain into a river, and a weekly league puts your streak against other students. Most retrieval and flashcard apps lose people in the first two weeks because the work is invisible and adherence is left to willpower. The tree-per-deck mechanic is the part that turns a five-minute habit into something you can see accumulating, which is what keeps it alive.

Are the questions any good, or is this just ChatGPT output?

Question quality matters here, because drilling bad questions degrades back into recognition practice, which is rereading with extra steps. Studyly runs a four-criterion rubric on every card (factual correctness, stem clarity, distractor plausibility, question-type coverage). On a held-out three-document eval it scored 81.3, against 78.0 for Unattle, 68.0 for Gauntlet, and 57.8 for Turbolearn. The methodology is published on the studyly.io quality page.

What if my exam is tomorrow and I have no time to build a habit?

Then the habit framing on this page does not apply, but drilling still beats rereading even from a single cramming session. The move is the same, just compressed: convert every deck you have left and drill instead of skim-rereading. The cramming-specific protocol is covered on the retrieval-beats-rereading cramming guide on this site.