Spaced Repetition Flashcards — Anki, Quizlet, RemNote, and Memory Research

Spaced repetition flashcards are one of the most well-researched study techniques in cognitive psychology. Per Cepeda et al.’s landmark 2006 meta-analysis of 254 spaced-practice studies, distributing practice across time produces 50-200% better long-term retention than massed practice (cramming). Combined with active retrieval testing effects (per Roediger and Karpicke), spaced flashcards represent some of the highest-leverage learning techniques available.

This article uses peer-reviewed cognitive psychology research (Cepeda meta-analysis, Roediger and Karpicke testing effect, Ebbinghaus forgetting curve), Anki documentation, and education researcher Andy Matuschak’s flashcard methodology to evaluate spaced repetition apps and techniques. Topics include the science behind algorithms, app comparison, card design principles, and realistic study plans.

For complementary content, see note-taking apps compared and language-learning apps data.

The cognitive science

Hermann Ebbinghaus’s 1885 forgetting curve research established that memory decay follows a predictable curve — most information fades within hours of learning if not reinforced. Modern cognitive psychology has refined this:

Without review, retention drops sharply: ~50% lost within 1 hour, ~70% lost within 24 hours, ~80% lost within 1 week.

With strategic review at increasing intervals (1 day, 3 days, 1 week, 1 month, 3 months), retention reaches 80%+ long-term with much less total study time than cramming.

The Cepeda et al. meta-analysis quantified this: optimal spacing is approximately 10-20% of the desired retention interval. For 1-week retention, review 1-2 days after learning. For 1-year retention, review every 1-2 months.

SuperMemo and SM-2

Piotr Wozniak’s SuperMemo research (1980s onwards) operationalized this into algorithms. The SM-2 algorithm (1987) and its derivatives became the foundation for modern spaced repetition apps. The algorithm tracks per-card review performance, adjusts intervals based on difficulty, and schedules optimal next-review dates.

Anki’s algorithm is a derivative of SM-2 with modifications. Modern apps like FSRS (Free Spaced Repetition Scheduler) optimize further using machine learning on individual learner data.

Anki — the serious tool

Anki is the established serious flashcard tool. Free on desktop and Android, $25 one-time on iOS (the developer’s main income source). Powers medical school study, language acquisition, and serious long-term memorization.

Anki’s strength is depth. The card creation interface supports cloze deletions, image occlusion (for anatomy), audio (for language), and complex card types. The shared deck ecosystem includes pre-made content for almost any major subject area.

The learning curve is meaningful — first 1-2 weeks of Anki use can be frustrating as you learn card design principles and workflow. The payoff is durable long-term retention with surprisingly little ongoing time investment.

Quizlet — the friendly default

Quizlet dominates student flashcard use in U.S. schools. The interface is simple, shared decks abundant, and study modes engaging (matching games, write mode, test mode).

Quizlet’s strength is accessibility — students start using it in 5 minutes. The shared deck ecosystem matches typical high school and college subjects well.

Weaknesses include less sophisticated spacing algorithm (despite “Quizlet Learn” mode), and recent emphasis on AI features that may distract from core study workflow.

Best for: high school and undergrad students with shorter study cycles (per-semester courses), shared class decks, casual review.

RemNote — note-taking + flashcards

RemNote combines note-taking with spaced repetition flashcards in one tool. Notes can be tagged as cards with cloze deletion or question-answer pairs, scheduled for review through SM-2 algorithm.

The integrated approach addresses one of Anki’s pain points: creating cards in a separate app away from your notes. With RemNote, notes-as-cards work in one workflow.

Best for: students who want unified notes-and-cards system, PKM practitioners, learners willing to commit to a single tool for notes and review.

Weakness: less mature than Anki or Quizlet, smaller user base, fewer shared decks.

Card design principles

Per Andy Matuschak’s research and SuperMemo principles, card quality matters more than card quantity:

Atomic — one fact per card. Bad: “What are the four humors and their associated personality types?” (4 facts in 1 card). Good: 4 separate cards, one per humor.

Specific — clear, single correct answer. Bad: “Explain the French Revolution” (multiple valid answers). Good: “In which year did the Bastille fall?” (1789).

Active production — fill-in or generate rather than recognize. Bad: “Which of these is the French word for ‘cat’? A) chat B) chien C) oiseau” (multiple choice = recognition). Good: “What is the French word for ‘cat’?” (active production).

Cloze deletion when possible — fill-in-the-blank for facts embedded in sentences (cloze deletion software hides one term such as the year 1789 from the sentence “The Bastille fell in 1789”). This preserves context that helps recall.

Context when needed — pure isolation can hurt memory. “(In the context of the French Revolution) when did the Bastille fall?”

Image and audio — multimedia cards outperform text-only for many subject areas (anatomy, music, language pronunciation).

Time management

Per Anki community data and student reports, realistic time budgets:

Casual learner (vocabulary, 50 cards/day new): 15-25 min/day for review and new card processing.

Serious language learner (20-30 new cards/day): 25-45 min/day.

Medical student (Step 1 prep, 100+ cards/day new): 60-120 min/day.

The pattern: more new cards added = more review time later. Consistency matters more than session length — daily 15 minutes outperforms weekly 2 hours for long-term retention.

The consistency trap: missing 3 days creates 3x the workload returning (3 days of accumulated reviews due). Easier to maintain daily habit than to recover from gaps.

Pre-made vs custom decks

Pre-made decks save creation time but have downsides:

Cards may not match your knowledge level (already-known facts wasted as cards, unknown facts missing). The cards aren’t yours — creating cards itself is learning. Card quality varies (community decks include poor-quality cards mixed with good ones).

Best practice: use pre-made decks as starting point. Mark already-known cards for deletion aggressively. Add custom cards for high-priority unknowns missed by the deck.

Notable shared decks: AnKing (medical school), Refold (language acquisition), AwesomeAnki (various subjects), Migaku decks (language).

Common mistakes

Per Anki community feedback:

Over-collection — adding 50+ new cards/day quickly creates unsustainable review load. Better: 10-20 new cards/day sustained for months.

Skipping difficult cards — using “Again” button vs muddling through builds long-term retention better than fake “Good” ratings.

Reviewing only when stressed — Anki works best as daily habit. Sporadic use loses the algorithm’s effectiveness.

Bad card design — vague, ambiguous, or non-atomic cards waste review time without building retention. Better to delete poor cards than struggle with them.

Bottom line

For serious long-term learning (language, medical knowledge, foundational disciplines): Anki is the strong choice. The learning curve is real but the depth and durability of retention justify the investment.

For casual student use (per-semester courses, shared class decks): Quizlet is sufficient and easier to start.

For unified notes-and-cards workflow: RemNote merits consideration.

The technique itself — spaced repetition with active retrieval — is one of the highest-leverage learning techniques available. Even modest daily practice (15-25 minutes) compounds into significant long-term retention.

For complementary reading, see note-taking apps compared, language-learning apps data, and the study methods category.