Spaced repetition is supported by evidence from laboratory studies, classrooms, and professional education. Across 29 studies, spaced retrieval produced a large advantage over massed retrieval for final retention, with an effect size of g = 0.74 in a 2021 meta-analysis. Results vary by subject, interval, assessment delay, and study design, so the statistics below should be read within their stated populations and time frames.
Contents
- What the meta-analyses find
- How intervals affect retention
- Why delayed retrieval matters
- Results in classrooms and college courses
- Evidence from medical and professional education
- Anki, engagement, and transfer
What the meta-analyses find
The broadest evidence in the supplied research comes from a 2021 meta-analytic review of spaced retrieval practice. Across 29 studies, spaced retrieval produced a large advantage over massed retrieval on final retention, measured as g = 0.74. The review included 39 effect sizes in the spaced-versus-massed comparison. A Meta-Analytic Review of the Benefit of Spacing out Retrieval Practice Episodes on Retention (2021).
The same review compared two ways to arrange spaced retrieval. Expanding schedules and uniform schedules did not differ significantly: g = 0.034. That comparison contained 54 effect sizes. This result does not mean that spacing is ineffective; it means the available evidence did not establish a reliable advantage for one of those two spacing patterns over the other in that review.
Mathematics research points in the same general direction, with smaller average effects. A 2025 meta-analysis found that spaced practice versus massed practice produced a small-to-medium overall effect of g = 0.28 across 27 studies and 53 effect sizes. When the material was isolated mathematics content, the effect was g = 0.43 across 10 studies and 27 effect sizes. For course-embedded mathematics instruction, the effect was g = 0.24 across 17 studies and 26 effect sizes. A Meta-Analytic Review of the Effectiveness of Spacing and Retrieval Practice for Mathematics Learning (2025).
That mathematics review also separated the effects of testing from those of restudying. Testing versus restudying had a weighted mean effect of g = 0.18 across 7 studies and 32 effect sizes, but its 95% interval crossed zero. The result therefore does not provide a statistically secure advantage for testing over restudying in that specific analysis.
How intervals affect retention
A vocabulary experiment published in 2011 illustrates why the length of the spacing interval can matter. On a final recall test one week later, the long-spacing condition produced 75% recall, compared with 64% for medium spacing and 49% for short spacing. Three massed retrievals produced 26% final recall, almost the same as one retrieval at 25%. One-time study produced only 1% final recall in the experiment. Spaced Retrieval: Absolute Spacing Enhances Learning Regardless of Relative Spacing (2011).
In that comparison, long spaced retrieval improved final recall by 200% relative to three massed retrievals. The long-spacing condition used mean lags of approximately 15, 30, and 45 intervening items for expanding repetitions. During learning, repeated-test retention was high in all three conditions: 97% with short spacing, 96% with medium spacing, and 90% with long spacing. The lower immediate performance in the long-spacing condition therefore coexisted with better delayed recall.
The experiment assigned 184 items to short expanding spacing, 185 to medium equal spacing, and 188 to long equal spacing. Relative spacing did not have a significant effect on final retention, with F < 1. Absolute spacing did have a significant effect, F(2,69) = 11.07, and explained 24% of partial variance in final recall, ηp² = .24. The authors’ distinction is important: the total amount of intervening material mattered in this experiment, while the proportional pattern of the intervals was not discernible as a separate influence.
| Condition or comparison | Quantified result | Measurement context |
|---|---|---|
| Long spacing | 75% recall | One-week final test |
| Medium spacing | 64% recall | One-week final test |
| Short spacing | 49% recall | One-week final test |
| Three massed retrievals | 26% recall | One-week final test |
| One retrieval | 25% recall | One-week final test |
| One-time study | 1% recall | Vocabulary experiment |
Response time showed a more limited relationship with learning. The response-time slope correlated negatively with final recall across subjects, r = −.20. Within subjects, response-time-slope correlations with recall averaged r = .04 and were not significant. These figures suggest that slower or faster responses cannot be treated as a simple universal proxy for later retention.
Why delayed retrieval matters
Spacing often looks less comfortable during learning because a delay can make retrieval harder. A 2018 review example reported nearly perfect immediate recall after massed practice, followed by about a 25% drop when five intervening items were inserted before retrieval. Yet spaced retrieval outperformed massed retrieval on final tests delayed by both 10 minutes and 2 days. Retrieval-Based Learning: A Decade of Progress (2018).
The best schedule is not established by every study. In one comparison, an expanding schedule with gaps of 2, 6, and 19 days produced a 3% advantage over uniform 9-, 9-, and 9-day gaps at a 56-day final test, but the difference was not significant. Another study compared 1-and-5-day relearning intervals with 3-and-3-day intervals and found no overall expanding-schedule superiority across retention tests at 15 minutes, 1 day, 7 days, and 35 days. At the 1-day and 7-day points, contracting intervals produced the best performance in that schedule comparison. Retrieval-Based Learning: A Decade of Progress (2018; studies discussed from 2014).
Timing also mattered in a study of two undergraduate physiology courses. Homework assigned 5 days after class improved end-of-semester topic scores compared with homework assigned 1 day after class. The study assessed weekly retention after 1 week and final-exam retention after delays of 4 to 13 weeks. One-week information retention did not differ between the 1-day and 5-day schedules in either course, showing how an interval can appear beneficial at a later assessment without changing an earlier one. Retrieval practice in the form of online homework improved information retention more when spaced 5 days rather than 1 day after class in two physiology courses (2018).
Results in classrooms and college courses
Applied classroom studies show that spacing can improve learning, but the size of the outcome depends on the course and test. In a 2023 financial-education experiment, mean final knowledge accuracy was 0.61 with spaced practice, 0.49 with massed practice, and 0.54 in a control condition. The practice-condition effect was significant, F(2,172) = 5.654, p = .004, with partial η² = .062. Spaced practice exceeded massed practice with d = .66 and exceeded controls with d = .39. Massed practice did not differ significantly from control practice, p = .197. Using Cognitive Science and Technology to Enhance Financial Education: The Effect of Spaced Retrieval Practice (2023).
Behavioral evidence helps explain why study schedules may fail to match recommended practice. A naturalistic 2021 undergraduate study logged spaced-study behavior from 185 students. Spaced quizzing predicted course performance and GPA more strongly than spaced textbook reading. Students used both a textbook tool and a quiz tool daily, but textbook use exceeded quiz-tool use. Many students quizzed themselves only 1 day per exam period, often near the deadline. These associations describe study behavior and outcomes; they do not by themselves prove that quizzing caused the performance differences. Students’ Spacing of Their Study (2021).
A medical lecture trial included 148 students and used three 15-minute instructional periods separated by two 5-minute intervals. Eight days after the lecture, spaced students averaged 33.8%, compared with 31.8% for traditional lectures. After baseline adjustment, the difference was not significant, F(1,104) = 0.566, p = .458. Making a Lecture Stick: the Effect of Spaced Instruction on Knowledge Retention in Medical Education (2021).
Evidence from medical and professional education
Some professional-education trials report larger differences. A 2024 randomized dental trial assigned 90 students equally to lecture-based control and mobile spaced-repetition groups. The spaced-repetition group had significantly higher retention than lecture controls at both 1 month and 3 months, with p ≤ .001 for each comparison. Effectiveness of spaced repetition learning using a mobile flashcard application among dental students: A randomized controlled trial (2024).
A 2025 paediatrics study randomized 90 final-year medical students into 45 intervention and 45 control students. Pre-test means were 11.42 for spaced-repetition students and 11.58 for controls, p = .0573. Post-test means were 16.244 and 11.89, respectively, with p < .0001. Implementation of a spaced-repetition approach to enhance undergraduate learning and engagement in paediatrics (2025).
The largest reported professional sample was a 2024 study of physicians and residents. It randomized 26,258 baseline completers across a control group and five spaced-repetition conditions. At quarter 6, learning performance was 58.03% with spaced repetition versus 43.20% without spacing, d = .62, p < .001. At quarter 10, knowledge transfer was 58.33% versus 52.39%, d = .26, p < .001. Double-spaced repetition produced 62.24% learning versus 51.83% with single spacing, d = .43, p < .001; transfer was 60.08% versus 55.72%, d = .20, p < .001. The Effect of Spaced Repetition on Learning and Knowledge Transfer in a Large Cohort of Practicing Physicians (2024).
Other results are more modest. In a 204-student histology study, spaced-application users answered about 901 items compared with 858 for controls, p = .73. The application group scored 11.8 versus 10.98 on a 15-point practical exam, p = .016. Efficiency of web application and spaced repetition algorithms as an aid in preparing to practical examination of histology (2019).
Anki, engagement, and transfer
Anki studies provide useful but often observational evidence about how learners use spaced repetition. In a first-year medical-school study, 26 of 29 respondents used Anki, and 25 valid database files were analyzed. Users scored 6.4% higher on Course I, 6.2% on Course II, 7.0% on Course III, and 12.9% on CBSE exams. A Cohort Study Assessing the Impact of Anki as a Spaced Repetition Tool on Academic Performance in Medical School (2023).
A separate 2025 Anki cohort had a mean CBSE exam score of 65.2%, with a standard deviation of 11.6%, across 36 students. Students above the mean mature-card count scored 71.5%, compared with 60.0% for students below the mean. Mature-card count correlated with CBSE scores at p = .002, while total study hours correlated at p = .013. Because these are cohort associations, the figures should not be interpreted as proof that card counts alone caused higher scores. Exploring the Impact of Spaced Repetition Through Anki Usage on Preclinical Exam Performance (2025).
Spacing effects also appear in other learning tasks. A 2019 pharmacy study found spaced standard retrieval was better than non-spaced retrieval on long-term retention, d = .43, while elaborated spaced retrieval reached d = .54. In a 2017 naming-rehabilitation study, spacing improved one-day retention with coefficient 0.71, SE 0.16, p < .001; the benefit remained at one-week follow-up with coefficient 0.45, SE 0.16, p = .006. Optimizing the Spacing of Retrieval Practice to Improve Pharmacy Students’ Learning of Drug Names (2019); Towards a Theory of Learning for Naming Rehabilitation: Retrieval Practice and Spacing Effects (2017).
Age and lesson format can alter the pattern. In an experiment comparing 1, 3, and 5 retrieval tests, repeated spaced testing improved retention over one spaced test for both young and older adults. Repeated massed testing improved retention only for young adults. A 2024 randomized nurse-anesthesia study assigned 64 students to spaced lectures or conventional 90-minute lectures; the spaced sessions were three 30-minute blocks separated by 10-minute intervals. Retrieval Practice and Spacing Effects in Young and Older Adults: An Examination of the Benefits of Desirable Difficulty (2015); The effect of spaced learning on the learning outcome and retention of nurse anesthesia students: a randomized-controlled study (2024).
The evidence base is heterogeneous across vocabulary, mathematics, financial education, and health-professional settings. Publication dates and study periods are reported by the cited sources, and several studies are correlational or retrospective. Results should therefore remain tied to the stated learners, materials, schedules, and retention delays rather than being generalized to every study task.