Retrieval practice means pulling what you have learned out of memory, with the notes shut, and only then checking. A brain dump on a blank page, a flashcard answered before it is turned, a past paper sat under exam conditions: each does more for memory a day or a week later than rereading. Researchers call the benefit the testing effect.
How much more depends on how you do it: checking answers, recalling rather than recognising, and not dropping what you already know. This guide gives the findings as the papers state them, with their limits, then the methods four universities recommend, and a plan.
EdCitation publishes this guide. The studies were read in the papers themselves, except three read only in their abstracts, which we name. EdCitation's Cite a source built every study reference from its DOI, Verify references checked the whole list, and Find sources is where to look for more. The plan near the end is ours, and untested.
What is retrieval practice?
Retrieval practice is any study activity in which you produce the answer from memory before you see it. Study-skills pages call it active recall; York's library guide says it makes the brain "pull information out" (University of York Library, n.d.).
The classic experiment, in which students who recalled a passage beat those who reread it after two days and a week, is set out in our leader guide, how to study effectively: what the research shows (Roediger & Karpicke, 2006). This guide covers how large the effect is, when it shrinks, and how to use it.
What does the research on the testing effect show?
The research shows that recalling beats restudying on later tests, by an amount that depends on feedback, the kind of test and the delay.
The laboratory evidence, pooled
Rowland (2014) pooled 159 effect sizes from 61 studies reported between 1975 and 2013, each comparing tested with restudied material. The average effect was g = 0.50 (95% CI [0.42, 0.58]), and 81% of the effects favoured testing. The moderators are the practical part:
- Feedback. Studies that showed the answer again after the test found g = 0.73; those without feedback found 0.39.
- Delay. With a final test a day or more later the effect was 0.69; within a day, 0.41.
- Success on the practice test. Without feedback, getting half or fewer right gave no reliable benefit (g = 0.03); above 75% it gave 0.56.
- Recall against recognition. Among studies with feedback or high practice scores, cued and free recall (0.72 and 0.81) beat recognition, identifying rather than producing the answer (0.36). Across all studies free recall and recognition both gave 0.29.
The paper states its limits: no classroom studies, samples almost all undergraduates, and larger effects in published than unpublished studies, a sign of possible publication bias.
Against a harder rival: concept mapping
Rereading is a weak rival, so Karpicke and Blunt (2011) set retrieval against concept mapping, an active task. Eighty undergraduates studied a science text; one group drew a concept map with the text in view, another recalled freely, reread and recalled again, in the same time. A week later, on factual and inference questions, the retrieval group scored 0.67 against 0.45 (d = 1.50). Concept mapping did no significantly better than extra reading.
A second experiment, with 120 students, held even when the final test was a concept map drawn from memory (d = 1.01): 101 did better after retrieval, yet 90 had expected concept mapping to do as well or better. A map drawn with the text closed, the authors add, is itself retrieval.
The result has been independently repeated. Camerer et al. (2018) replicated 21 experiments from Nature and Science; 13 held, with effects about half the original size on average. Karpicke and Blunt's held with 49 participants, at an effect about 64% of the original's.
In real classrooms
Agarwal et al. (2021) screened nearly 2,000 abstracts and coded 50 classroom experiments. Their abstract reports 49 effect sizes from 5,374 students, 57% of them medium or large, with benefits across education levels, subjects, test delays and formats; only 6% came from what the authors call non-WEIRD countries. We read the abstract only; the full text sat behind a bot check. Adesope and colleagues' 2017 meta-analysis, discussed in the leader guide, reached the same verdict against restudying (Adesope et al., 2017).
Does it help with new questions?
Retrieval practice helps with new questions too, within limits. Pan and Rickard (2018) pooled 192 effects from 122 experiments (N = 10,396) in which the final test asked something different from the practice. Transfer averaged d = 0.40. It was largest across test formats and to application and inference questions, and among the weakest to material studied but never tested. After correcting for publication bias, the predicted transfer fell sharply, often to none when three conditions were all missing: a practice answer of the same kind as the final one, elaborated retrieval, and high practice-test success.
Where researchers disagree
Researchers disagree about complex material. Van Gog and Sweller argued that the testing effect shrinks as material grows more complex and may vanish when it is very complex (van Gog & Sweller, 2015; we read the abstract). Karpicke and Aue (2015) replied that complexity was never measured or varied in the studies cited, that studies with complex material were left out, and that the failures used isolated words or immediate, massed recall; the worked-example data, they say, still show a small benefit. Our reading: the effect is surest for facts and texts, smaller for problem-solving from worked examples.
Which active recall techniques work, and how do you do each?
The techniques that work all make you produce the answer before you see it, then check it.
| Method | How to do it | Evidence | Source |
|---|---|---|---|
| Brain dump (blank page, blurting) | Notes shut, write all you know on a topic for about five minutes, check, restudy the gaps | Free recall beat concept mapping with the text open, a week later | Karpicke and Blunt, 2011; University of Dundee, n.d. |
| Practice questions you write | Turn learning objectives and lecture material into questions; answer in writing, then check | Recall gave about twice recognition's effect where practice had feedback | Rowland, 2014; University of York Library, n.d. |
| Flashcards | Answer before turning; keep known cards in the deck; later, show the answer and give the question | Dropping items after one right answer cut recall a week later from about 80% to about a third | Karpicke and Roediger, 2008; University of Dundee, n.d. |
| Past papers and mock exams | Timed, closed book, no answer key, then mark | Classroom quizzing helped across levels and subjects | Agarwal et al., 2021; Cornell University Learning Strategies Center, n.d.-b |
| Concept map from memory | Draw the map with the text closed, then compare | Called retrieval by the authors; not tested as a study method | Karpicke and Blunt, 2011 |
| Multiple-choice quizzes | Use them, but add questions you must answer in words | Recognition gave smaller effects than recall | Rowland, 2014 |
| Reading the studies yourself | Search by the claim, then check the list | All 15 of this guide's references verified, the one PDF among them | EdCitation's Find sources and Verify references |
We read the universities' pages on 27 September 2026; York's guide was last updated on 2 September 2026, Dundee's on 24 September.
Brain dumps
Dundee's guide gives the method (a blank sheet, about five minutes, then the notes, then time on what you missed) and warns against glancing at a gap and deciding "oh, I actually know that" (University of Dundee, n.d.). Cornell University's Learning Strategies Center calls it Blank Page Testing and allows flow charts and diagrams as well as prose (Cornell University Learning Strategies Center, n.d.-a). York St John's fact sheet ends with repeating the whole process (York St John University, n.d.).
Flashcards
A flashcard is retrieval practice only if you answer before you turn it; Dundee says so outright. York's guide lists the drawbacks: cards take time to make, answers can be anticipated so the practice turns passive, and facts learned card by card can lose their connections. Karpicke and Roediger (2008) show why known cards stay in: once students could recall a Swahili word, dropping it from further tests cut recall a week later from about 80% to 36%, while dropping it from further study cost nothing. York St John's three-box Leitner method tests known cards less often but keeps them in, which fits that finding.
Practice questions and past papers
Cornell advises sitting a practice test in the time allowed, with only the resources the real exam permits, once at the start of revision as a baseline and again after (Cornell University Learning Strategies Center, n.d.-b). Dundee's warning matches Pan and Rickard's finding: use question banks, but not only them, or you will know the questions and have gaps elsewhere. Without past papers, York suggests writing questions from the textbook, the lectures and the learning objectives. Dundee also suggests asking a chatbot to quiz you; check its answers against your notes, since chatbots state wrong things confidently, as why AI tools invent references shows.
How do you get the most from retrieval practice?
You get the most from retrieval practice by checking every answer, recalling rather than recognising, and retesting what you already know. From the studies above:
- Check every answer: feedback was among Rowland's strongest moderators.
- Keep going after a right answer: Karpicke and Roediger's students forgot most of what they stopped retrieving.
- Write or say it: with feedback, recall beat recognition about two to one.
- Test the whole syllabus: untested material transferred among the least.
- Trust it over your feeling: three quarters of Karpicke and Blunt's students misjudged which method worked.
What does a week of retrieval practice look like?
A week of retrieval practice puts a short recall session after each lecture and a mixed test at the end. The plan is our own synthesis of the findings; nobody has tested it as a package.
- The same day as each lecture or reading, do a five-minute brain dump. Check it against the notes and mark what you missed.
- Turn every gap and every learning objective into a question, on a card or in a list.
- Midweek, answer every card from memory before turning it, including the ones you got right last time.
- At the end of the week, answer closed-book short-answer questions from the whole week, mark them, restudy what you got wrong, and retest it the next day.
- If you score half or less, restudy before testing again, our reading of Rowland's success-rate finding.
- In the weeks before an exam, add a past paper under timed, closed-book conditions, then write questions on topics the papers never asked.
- For an essay rather than an exam, look up a source when a point needs one. Search EdCitation's Find sources and build the reference in Cite a source as you write the note.
Can EdCitation help you study?
For the research behind a study tip, EdCitation is the best tool: its references come from the publisher's own record rather than being composed, which no chatbot can promise. It never writes a word of a student's work and will not quiz you; it finds, cites and verifies the studies.
Find the studies
Find sources searches about 300 million published works. On 27 September 2026, a search for "retrieval practice classroom learning", with no filter, returned 4,464,904 works. The first was a McDaniel record from 2009 titled "Improving Student Learning: Importing Retrieval Practice Effects Into The Classroom", in PsycEXTRA Dataset, cited 0 times; the first ten included Agarwal, Bain and Chamberlain's 2012 paper in Educational Psychology Review, cited 103 times.
Cite them
Paste a study's DOI into Cite a source (a title, PubMed ID, ISBN or link also works) and it returns the reference in any of six styles, from APA 7 to Vancouver, with a retraction screen. From 10.1037/a0037559 it returned, correctly:
Rowland, C. A. (2014). The effect of testing versus restudy on retention: A meta-analytic review of the testing effect. Psychological Bulletin, 140(6), 1432–1463. https://doi.org/10.1037/a0037559
From 10.1126/science.1152408 it returned "Karpicke, J. D., & Roediger, H. L., III. (2008). The critical importance of retrieval for learning. Science, 319(5865), 966–968.", with the title in the sentence case APA 7 asks for (title case or sentence case explains why), as it did for Karpicke and Blunt's, Agarwal's and both 2015 titles. Pan and Rickard's entry, and Camerer's with its 24 authors and "Nature and Science" kept capitalised, came back right too. No retraction notice came back for any of them.
Verify the list
Verify references takes a pasted list or a whole paper, looks each entry up in the publishers' records, flags retractions, and never turns "could not check" into "not found". This guide's 15 references came back 15 verified, none doubtful, not found or retracted. The York St John fact sheet, a PDF, was read too, with the note "The address opens a PDF whose first pages carry the entry's title." Adesope and colleagues' entry arrived with a pointer to a correction published in 2017.
What it costs
All three are free, with no account; Pricing lists Pro and Max.
Quick questions
Is active recall the same as retrieval practice?
Active recall and retrieval practice are the same thing: study-skills pages use the first name, researchers the second, and both mean producing the answer from memory before checking.
Does retrieval practice work if I get the answers wrong?
Retrieval practice works when you check your answers. In Rowland's meta-analysis, practice without feedback on which learners got half or fewer right gave no reliable benefit; practice with feedback gave the largest effects.
Are multiple-choice quizzes good retrieval practice?
Multiple-choice quizzes help, but less. Multiple choice is recognition, which gave about half the effect of recall in Rowland's studies with feedback or high practice scores. Add questions you answer in your own words.
Should I stop testing myself on flashcards I know?
Keep flashcards you know in the deck. In Karpicke and Roediger's experiment, students who stopped testing words after one correct recall remembered about a third of them a week later, against about 80% for those who kept testing.
Where can I read the research on the testing effect?
Search EdCitation's Find sources for "testing effect" or "retrieval practice", then check the studies you pick in Verify references before you cite them.
References
- Adesope, O. O., Trevisan, D. A., & Sundararajan, N. (2017). Rethinking the use of tests: A meta-analysis of practice testing. Review of Educational Research, 87(3), 659–701. https://doi.org/10.3102/0034654316689306
- Agarwal, P. K., Nunes, L. D., & Blunt, J. R. (2021). Retrieval practice consistently benefits student learning: A systematic review of applied research in schools and classrooms. Educational Psychology Review, 33(4), 1409–1453. https://doi.org/10.1007/s10648-021-09595-9
- Camerer, C. F., Dreber, A., Holzmeister, F., Ho, T.-H., Huber, J., Johannesson, M., Kirchler, M., Nave, G., Nosek, B. A., Pfeiffer, T., Altmejd, A., Buttrick, N., Chan, T., Chen, Y., Forsell, E., Gampa, A., Heikensten, E., Hummer, L., Imai, T., … Wu, H. (2018). Evaluating the replicability of social science experiments in Nature and Science between 2010 and 2015. Nature Human Behaviour, 2(9), 637–644. https://doi.org/10.1038/s41562-018-0399-z
- Cornell University Learning Strategies Center. (n.d.-a). Effective study strategies. Retrieved September 27, 2026, from https://lsc.cornell.edu/how-to-study/studying-for-and-taking-exams/effective-study-strategies/
- Cornell University Learning Strategies Center. (n.d.-b). What to do with practice exams. Retrieved September 27, 2026, from https://lsc.cornell.edu/how-to-study/studying-for-and-taking-exams/what-to-do-with-practice-exams/
- Karpicke, J. D., & Aue, W. R. (2015). The testing effect is alive and well with complex materials. Educational Psychology Review, 27(2), 317–326. https://doi.org/10.1007/s10648-015-9309-3
- Karpicke, J. D., & Blunt, J. R. (2011). Retrieval practice produces more learning than elaborative studying with concept mapping. Science, 331(6018), 772–775. https://doi.org/10.1126/science.1199327
- Karpicke, J. D., & Roediger, H. L., III. (2008). The critical importance of retrieval for learning. Science, 319(5865), 966–968. https://doi.org/10.1126/science.1152408
- Pan, S. C., & Rickard, T. C. (2018). Transfer of test-enhanced learning: Meta-analytic review and synthesis. Psychological Bulletin, 144(7), 710–756. https://doi.org/10.1037/bul0000151
- Roediger, H. L., III, & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255. https://doi.org/10.1111/j.1467-9280.2006.01693.x
- Rowland, C. A. (2014). The effect of testing versus restudy on retention: A meta-analytic review of the testing effect. Psychological Bulletin, 140(6), 1432–1463. https://doi.org/10.1037/a0037559
- University of Dundee. (n.d.). Learning and revision. Academic Skills. Retrieved September 27, 2026, from https://libguides.dundee.ac.uk/academicskills/learningandrevision
- University of York Library. (n.d.). Active recall. Study and revision: A practical guide. Retrieved September 27, 2026, from https://subjectguides.york.ac.uk/study-revision/active-recall
- van Gog, T., & Sweller, J. (2015). Not new, but nearly forgotten: The testing effect decreases or even disappears as the complexity of learning materials increases. Educational Psychology Review, 27(2), 247–264. https://doi.org/10.1007/s10648-015-9310-x
- York St John University. (n.d.). Exam revision: Memory and recall [Fact sheet]. Retrieved September 27, 2026, from https://www.yorksj.ac.uk/media/content-assets/study-skills/maths-and-statistics/core-skills/Exam-Revision-Memory-Recall-Factsheet.pdf