The Science of Spaced Repetition: How It Works and Why

In a 2025 randomized study at a medical college in India, 90 final-year students were split into two groups studying the same paediatrics material. One group reviewed flashcards at increasing intervals: 1, 3, 7, 14, and 28 days. The other group studied the traditional way: lectures, notes, textbooks. Both groups started at nearly identical pre-test scores, around 11.5 out of 20.

After the study period, the traditional group had moved from 11.58 to 11.89, a change so small it wasn't statistically significant. The spaced repetition group moved from 11.42 to 16.24, a jump that was significant at p<0.0001. Same material, same students, same time investment. The only variable was when the review happened.

This is not an isolated finding. It is one data point in roughly a century of consistent research on why spacing out review sessions produces dramatically better retention than cramming, and why almost every well-designed learning system, from medical school curricula to language apps to AI-generated courses, builds spacing into its core structure.

Key Takeaways

• A 2025 randomized study found spaced repetition raised post-test scores by 4.8 points (11.42 to 16.24) while traditional study moved scores by 0.3 points over the same period (Vagha et al., Frontiers in Medicine, 2025)

• Medical students using Anki, a spaced repetition app, scored 6.2 to 10.7% higher on standardized exams than students using traditional study methods (Gilbert et al., 2023)

• The mechanism is grounded in Ebbinghaus's 1885 forgetting curve and confirmed by Cepeda et al.'s 2006 meta-analysis of distributed practice

• Optimal spacing intervals expand over time: roughly 1 day, then 3, then 7, then 14, then 30, adjusting based on how well you recall each review

• AI-generated courses can implement spaced retrieval automatically through built-in quizzes, removing the manual deck-building overhead that limits tools like Anki

What Is Spaced Repetition?

Spaced repetition is a learning technique that schedules review sessions at increasing intervals rather than all at once. Instead of reviewing material five times in one sitting (massed practice, commonly known as cramming), you review it once today, again in three days, again in a week, again in two weeks, and again in a month (distributed practice).

The intervals expand because each successful recall makes the memory more durable, which means you can wait longer before the next review without losing the information. This is the entire mechanism: review just before you would forget, and the gap between reviews grows over time as the memory strengthens.

This is different from simple repetition. Repeating something five times in a row produces weaker retention than the same five repetitions spread across days or weeks, even though the total exposure time is identical. The spacing itself, not just the repetition, is what produces the effect.

Why Does Spacing Work at All?

The foundational science here is more than a century old. Hermann Ebbinghaus, a German psychologist, ran a series of experiments on himself in 1885, memorizing nonsense syllables and tracking how quickly he forgot them. He found that without review, humans lose roughly half of newly learned information within an hour and the majority within a day. This forgetting curve has been replicated extensively over the following 140 years, and the full research breakdown covers the mechanism in detail.

The critical second half of Ebbinghaus's work, often overlooked, is what happens when you review material before it's fully forgotten. Each successful review flattens the curve. The first review might only extend retention by a few days. The second review, done at the right moment, extends it further. By the fourth or fifth correctly-timed review, the curve becomes nearly flat: the information has moved from fragile short-term memory into durable long-term storage.

Spaced repetition is the direct practical application of this finding. It schedules reviews at the point where the forgetting curve is about to drop steeply, catching the memory right before it would otherwise be lost.

What Does the Research Show?

The 2025 paediatrics study described above is a strong recent example, but the evidence base goes much deeper.

Cepeda et al. (2006): The Foundational Meta-Analysis

Cepeda et al. (2006) conducted a meta-analysis of distributed practice research and confirmed that spaced study produces meaningfully better long-term retention than massed study at the same total study time. This is one of the most replicated findings in cognitive psychology, holding across age groups, content types, and time scales.

Gilbert et al. (2023): Anki and Medical Exam Scores

Gilbert et al. (2023), studying medical students using Anki, a popular spaced repetition flashcard app, found that students using the spaced repetition tool scored 6.2 to 10.7 percentage points higher on standardized exams than students relying on traditional study methods.

Durrani et al. (2024): Independent Replication in Pakistan

This finding has been independently supported. Durrani et al. (2024), studying Pakistani medical students, found a statistically significant improvement in clinical problem-solving test performance among spaced repetition users, with a mean score increase of 2.93 points (p<0.01).

Kerfoot et al. (2007): Early Randomized Controlled Trial

Kerfoot et al. (2007), in one of the earlier randomized controlled trials on this topic, found that spaced education significantly improved retention of clinical knowledge among medical students compared to standard instruction, in a study design that has since been replicated across multiple medical specialties.

Tabibian et al. (2019): Computational Optimization

Tabibian et al. (2019), published in the Proceedings of the National Academy of Sciences, took the research a step further by developing a computational framework for optimizing spaced repetition schedules specifically for individual learners, rather than using fixed intervals for everyone. This work underpins the algorithm used by most modern spaced repetition software, including Anki's SM-2 algorithm and its successors.

What the data shows across studies

Sources: Frontiers in Medicine (2025), Medical Science Educator (2023), BMC Medical Education (2024), Psychological Bulletin (2006)

What Is the Optimal Spacing Schedule?

The exact intervals matter less than the principle of expansion, but research provides useful starting points.

A widely cited and well-tested schedule looks like: review after 1 day, then 3 days, then 7 days, then 14 days, then 30 days. Each interval roughly doubles, though modern algorithms adjust this dynamically based on individual performance rather than using a fixed sequence for everyone.

The 2025 paediatrics study used exactly this kind of escalating schedule (1, 3, 7, 14, and 28 days) and found it produced the dramatic score improvement described above. The interval doesn't need to be precisely optimized to work. What matters is that the gap expands over successive reviews and that review happens before the material is fully forgotten, not after.

If a review happens too early, before any meaningful forgetting has occurred, the session wastes time relearning what's still fresh. If a review happens too late, after the information has fully decayed, you're essentially relearning it from scratch rather than reinforcing an existing memory trace. The sweet spot, sometimes called the point of "desirable difficulty," is reviewing material right as it starts to feel effortful to recall.

Why Do Spaced Repetition Apps Have a Friction Problem?

Anki is the most established spaced repetition tool and the one most of the research above directly studied. It is also, by most users' own accounts, high-friction to start with.

To use Anki effectively, you need to create your own flashcard decks, or find and adapt community decks that match what you're studying. This means the spacing algorithm only works as well as the content you feed it. Poorly worded cards, cards covering the wrong level of detail, or gaps in deck coverage all reduce the system's effectiveness, and building good decks takes real time and skill.

This friction is well-documented. McConnery et al. (2021), studying spaced repetition adoption in a paediatrics curriculum, identified time constraints and inconsistent adherence as significant barriers to independent use, even when students recognized the method's value.

The constraint isn't the spacing algorithm. The algorithm works. The constraint is the manual labor of content creation that sits in front of it.

Spaced Repetition Without the Manual Overhead

This is where AI-generated learning tools change the equation in a way the research above couldn't anticipate, since most of these studies predate widely capable AI course generation.

Morso generates a structured course on any topic in about 30 seconds, with quizzes built into every lesson. The content creation step that consumes most of the friction in Anki-style spaced repetition is handled automatically. You don't build the deck. The course structure itself implements retrieval practice through quizzes embedded at each stage, and returning to a multi-lesson course across several days naturally creates the spacing intervals the research recommends.

This doesn't replace what Anki does for learners who specifically need to drill a defined, finite body of information they already know precisely (anatomical structures, vocabulary lists, dosage tables). For that use case, a dedicated spaced repetition system built around your own content remains the better tool.

But for the much larger population of self-directed learners exploring new topics, where the bottleneck isn't reviewing known material but generating well-structured content to review in the first place, AI-generated courses remove the step that historically kept most people from using spaced repetition consistently. The forgetting curve research explains why this matters: information you never get a structured chance to review is information you will lose regardless of how good the algorithm scheduling that review might be.

For a deeper comparison of how AI-generated and traditional flashcard-based spaced repetition compare, see best spaced repetition apps in 2026.

How Do You Apply Spaced Repetition Practically?

A few principles translate the research into something usable day to day.

Review before you fully forget, not after. The point of mild difficulty in recall, where you have to work a little to remember, is the optimal review moment. If recall feels effortless, you're reviewing too early. If you can't recall it at all, you've waited too long and are relearning from scratch.

Use active recall during review, not re-reading. Retrieval practice, actually trying to remember the answer before checking, produces substantially stronger retention than passively re-reading the material. This is true whether you're using physical flashcards, a spaced repetition app, or quizzes embedded in a structured course like those in Morso's AI-generated lessons.

Start with shorter intervals for harder material. Complex or unfamiliar concepts decay faster initially and benefit from a tighter first review (closer to 1 day rather than 3). Material you already find easy can tolerate longer initial gaps.

Let the schedule expand automatically. Whether using software with an adaptive algorithm or simply tracking review dates manually, the interval between reviews should grow each time you successfully recall the material. A fixed schedule that never expands wastes time on material you've already consolidated.

The Honest Limitations

Spaced repetition is not a universal solution, and the research is clear about where it falls short.

It is most effective for discrete factual and conceptual knowledge: vocabulary, definitions, clinical facts, historical dates, formulas. It is less directly useful for complex, integrated skills like writing, problem-solving in novel situations, or creative work, where the challenge isn't remembering a fact but applying judgment across many interacting variables.

McConnery et al.'s findings on adherence also matter. Even a perfectly designed spacing schedule fails if you don't actually do the reviews. The system works only as well as your consistency in using it, which is part of why removing friction from the content creation and review process (the problem AI-generated tools address) matters as much as the underlying algorithm. For more on how AI-generated and traditional formats compare overall, see microlearning vs traditional learning.

Sources

1. Vagha, K. et al. "Implementation of a spaced-repetition approach to enhance undergraduate learning and engagement in paediatrics." Frontiers in Medicine, 12:1601614. July 2025. https://doi.org/10.3389/fmed.2025.1601614

2. Cepeda, N.J., Pashler, H., Vul, E., Wixted, J.T., Rohrer, D. "Distributed practice in verbal recall tasks: A review and quantitative synthesis." Psychological Bulletin, 132(3):354-380. 2006. https://doi.org/10.1037/0033-2909.132.3.354

3. Gilbert, M.M. et al. "A cohort study assessing the impact of Anki as a spaced repetition tool on academic performance in medical school." Medical Science Educator, 33:955-962. 2023.

4. Durrani, S.F. et al. "Effectiveness of spaced repetition for clinical problem solving amongst undergraduate medical students studying paediatrics in Pakistan." BMC Medical Education, 24:676. 2024.

5. Tabibian, B., Upadhyay, U., De, A., Zarezade, A., Schölkopf, B., Gomez-Rodriguez, M. "Enhancing human learning via spaced repetition optimization." Proceedings of the National Academy of Sciences, 116:3988-3993. 2019. https://doi.org/10.1073/pnas.1815156116

6. Kerfoot, B.P., DeWolf, W.C., Masser, B.A., Church, P.A., Federman, D.D. "Spaced education improves the retention of clinical knowledge by medical students: a randomised controlled trial." Medical Education, 41:23-31. 2007.

7. McConnery, J.R., Bassilious, E., Ngo, Q.N. "Engagement and learning in an electronic spaced repetition curriculum companion for a paediatrics academic half-day curriculum." Perspectives on Medical Education, 10:369-372. 2021.

8. Ebbinghaus, H. Über das Gedächtnis (On Memory). 1885.