Spaced Repetition Research: What the Science Actually Shows (2026)
The Original Discovery: Ebbinghaus (1885)
The story of spaced repetition research begins with Hermann Ebbinghaus, a German psychologist who spent years memorizing nonsense syllables and testing his own memory at different intervals. His 1885 monograph Über das Gedächtnis (On Memory) documented two findings that have been replicated ever since:
The Forgetting Curve: Memory decays predictably after learning. Ebbinghaus found he forgot approximately 40% of new material within 20 minutes, 67% within a day, and nearly 80% within a month — without review.
The Spacing Effect: The same amount of study time distributed across multiple sessions produces far better long-term retention than the same time spent in a single session. Reviewing material after a delay is more effective than reviewing it immediately.
These findings, established in 1885, remain among the most replicated results in psychological research.
The Testing Effect: Roediger and Karpicke (2006)
The most cited modern study on retrieval practice is Roediger and Karpicke's 2006 paper "Test-Enhanced Learning: Taking Memory Tests Improves Long-Term Retention," published in Psychological Science.
The study: Students read prose passages and were assigned to one of three conditions:
- Study the passage four times (SSSS)
- Study three times, then take a practice test (SSST)
- Study once, then take a practice test three times (STTT)
The results:
- One week later, the STTT group (mostly testing) retained significantly more than the SSSS group (mostly studying)
- The difference was not marginal — the testing group retained approximately 80% compared to 36% for the re-study group
The key insight: Attempting to retrieve information from memory — even unsuccessfully — strengthens the memory trace more than re-reading the same content. The retrieval attempt, not the review, is the mechanism.
This is why spaced repetition produces such strong results: it forces retrieval practice at each review session.
Distributed Practice: The Cepeda et al. Meta-Analysis (2006)
Cepeda and colleagues published a major meta-analysis in Psychological Bulletin reviewing 254 studies on spaced versus massed practice (cramming).
Key findings:
- Distributed practice produced better retention than massed practice in 259 of 317 experimental comparisons
- The optimal spacing interval — the gap between study sessions — depends on the retention interval (how long you need to remember the material)
- For information needed after one week: optimal gap is roughly one day
- For information needed after a year: optimal gap is roughly a month
The lag effect: The meta-analysis confirmed that longer gaps between study sessions produce better long-term retention, up to a point. The optimal gap is approximately 10-20% of the retention interval — meaning if you need to remember something for a year, you should be reviewing it approximately monthly.
This is the mathematical basis for spaced repetition algorithms like Anki's FSRS.
The Optimal Spacing Algorithm: Wozniak and SuperMemo (1987-2006)
Piotr Wozniak began developing the SuperMemo algorithm in 1987, attempting to mathematically optimize review intervals based on Ebbinghaus's forgetting curve.
The SM-2 algorithm (1987) introduced the concept of exponentially increasing intervals: review a card when you get it right, and the next interval roughly doubles. Review less frequently as your memory strengthens, more frequently when it weakens.
The FSRS algorithm (Free Spaced Repetition Scheduler), developed more recently and now the default in Anki, represents the current state of the art. FSRS models memory as having two components:
- Stability — how long memory will persist before falling below a retention threshold
- Retrievability — the current probability of successful recall
Rather than fixed intervals, FSRS calculates the exact interval needed to maintain a target retention rate (typically 90%) for each individual card, based on your personal history with that card.
Research validation: The FSRS algorithm has been validated on large datasets of actual Anki usage data, consistently outperforming previous algorithms in predicted vs. actual retention.
Spacing vs. Interleaving: Rohrer and Taylor (2007, 2010)
Rohrer and Taylor investigated whether interleaving — mixing different topics within a study session — produces better retention than blocking (studying one topic at a time).
Finding: Interleaved practice produced significantly better performance on delayed tests, even when blocked practice performed better on immediate tests.
Why interleaving works: Blocked practice allows students to fall into routine patterns without thinking. Interleaved practice forces the brain to constantly identify which concept or strategy applies to each problem — the same cognitive demand as a real exam.
Practical implication: Study multiple subjects within a session rather than completing one subject entirely before moving to the next. Mix problem types within a problem set rather than completing all of one type before moving to another.
Desirable Difficulties: Bjork (1994, 2011)
Robert Bjork introduced the concept of "desirable difficulties" — learning conditions that make studying feel harder in the short term but produce better long-term retention.
The core insight: the conditions that feel most efficient during learning are often the conditions that produce the worst retention. And the conditions that feel inefficient — retrieval practice, spacing, interleaving — produce the best retention.
Why desirable difficulties work: When learning is easy, it signals that material is already in memory. When it's difficult — when you have to struggle to retrieve something — the act of retrieval strengthens the memory more. The difficulty is the mechanism, not an obstacle to learning.
This explains why students who re-read their notes feel like they know the material (it's familiar) but perform poorly on exams (familiarity isn't recall).
Effect Sizes: How Big Is the Benefit?
For context, effect sizes in educational research are typically classified as:
- Small: d = 0.2
- Medium: d = 0.5
- Large: d = 0.8
Spaced versus massed practice: Average effect size approximately d = 0.6 (medium-large) in Cepeda et al. meta-analysis, with significantly larger effects at longer retention intervals.
Testing versus re-reading: Roediger and Karpicke (2006) — approximately 44 percentage points difference (80% vs. 36%) after one week. Translated to effect size, approximately d = 1.0+ (large).
Interleaving versus blocking: Rohrer et al. studies show approximately 43% better performance on delayed tests for interleaved practice.
These are large effects by educational research standards, and they're consistent across studies, populations, and content domains.
Recent Research: AI and Adaptive Spacing (2020-2026)
Recent research has investigated whether AI-adaptive systems can further optimize spaced repetition beyond fixed algorithms.
Key findings from recent work:
- Adaptive systems that update spacing intervals based on individual performance data consistently outperform fixed-interval systems
- The FSRS algorithm used in Anki shows 10-15% improvement in retention efficiency compared to older algorithms in head-to-head comparisons on large datasets
- Machine learning approaches to predicting forgetting curves show promise but haven't yet demonstrated consistent superiority over FSRS in applied settings
The practical implication: Current best practice is using FSRS-based spaced repetition (Anki with FSRS enabled) combined with regular retrieval practice. More sophisticated AI-adaptive systems are in development but haven't yet displaced this as the evidence-based standard.
What the Research Means for How You Should Study
1. Test yourself more than you review The testing effect is one of the largest and most replicated findings in learning science. Replace re-reading with self-testing wherever possible.
2. Distribute study over time A single 90-minute session produces far worse retention than three 30-minute sessions across three days for the same total study time.
3. Use a spaced repetition system for memorization For content requiring long-term retention (medical school, language learning, professional certifications), a properly configured spaced repetition system (Anki with FSRS) implements the research evidence automatically.
4. Interleave subjects and problem types Mix different topics within study sessions rather than completing one topic entirely before starting another.
5. Embrace difficulty If studying feels easy, you're probably not learning effectively. The retrieval difficulty is the learning mechanism — don't shortcut it.
Key Papers for Further Reading
- Ebbinghaus, H. (1885). Über das Gedächtnis. Leipzig: Duncker & Humblot.
- Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning. Psychological Science, 17(3), 249-255.
- Cepeda, N. J., et al. (2006). Distributed practice in verbal recall tasks. Psychological Bulletin, 132(3), 354-380.
- Rohrer, D., & Taylor, K. (2007). The shuffling of mathematics problems improves learning. Instructional Science, 35(6), 481-498.
- Bjork, R. A. (1994). Memory and metamemory considerations in the training of human beings. In J. Metcalfe & A. Shimamura (Eds.), Metacognition.
- Ye, Z., et al. (2022). A New Algorithm for Scheduling Reviews. arXiv preprint. (FSRS development paper)
Frequently Asked Questions
Is spaced repetition just for memorization? Spaced repetition is most effective for content requiring verbatim recall — vocabulary, formulas, definitions, dates. For conceptual understanding (why something works, how to apply it), active retrieval practice through problem-solving and elaborative questioning is more appropriate. Most subjects benefit from both.
How much faster is spaced repetition than normal studying? Studies consistently show 10-30% better performance on immediate tests and 50-100%+ better performance on delayed tests with spaced practice compared to massed practice for the same total study time. Alternatively, spaced repetition can produce equivalent long-term retention with significantly less total study time.
Does spaced repetition work for everyone? Yes — the spacing effect is one of the most robust findings in cognitive psychology, replicated across ages, cultures, content domains, and ability levels. The magnitude varies but the direction of the effect is consistent.
What's the minimum effective spacing interval? For material needed a week later, reviewing after 24 hours produces significantly better retention than reviewing immediately. Even a one-night delay between study sessions produces measurable benefit over massed practice.
Does the research support using Anki specifically? The research supports spaced repetition and retrieval practice as learning principles. Anki is the most widely available implementation of these principles with a well-validated algorithm. The principles work regardless of implementation — a hand-made flashcard system following spacing principles will outperform cramming from notes.
Want to test whether you've retained the concepts in this article? Try the How to Learn Effectively Quiz — it's built around the same research principles described here.