What Is the 2357 Method?
The 2357 method (also written 2-3-5-7) is a spaced repetition schedule for revision. You learn a topic on day 1, then review it on days 2, 3, 5 and 7, so every topic gets four reviews in its first week. The gaps between sessions start at one day and grow to two.
Does it work? Mostly, yes, if you make each session a test and keep going once the schedule ends. The numbers are a rule of thumb from revision guides rather than a schedule tested in a lab, but they build in spacing, one of only two study techniques a major 2013 review rated "high utility". The other, practice testing, comes down to what you do in each session.
"Space out your revision" is advice nobody can act on at 9pm on a school night. Four dates in a planner are something you can act on.
Before you start:
- The numbers get read two ways. Some guides count them forward from the day you learn a topic. Others, including Birmingham City University's revision pages and The Complete University Guide, count them back from your exam date.
- The schedule only tells you when. What you do on each review day matters more. Pulling information out of memory works better than putting it back in front of your eyes.
The method pairs well with the blurting method, another UK revision staple: 2357 picks the days, and a blurt is what you do on each one.
Two Versions: The 7-Day Cycle and the Exam Countdown
Version 1: the 7-day cycle. Read the numbers as day numbers. Day 1 is when you first meet the topic, whether that's in class, in a textbook or in a lecture video. You review it on days 2, 3, 5 and 7 (that's 1, 2, 4 and 6 days after you learned it), and the cycle is done. This version suits term time, when new material arrives every week and there's no exam date yet.
Version 2: the exam countdown. Birmingham City University's revision guide starts from the exam instead. Plan a session the day before the exam, then count back two days and plan another, then three days, then five, then seven. The Complete University Guide gives the same recipe and adds a rule: make the session the day before each exam a past paper. Both note that the plan starts 18 days out, because 1 + 2 + 3 + 5 + 7 = 18.
Side by side, the two versions look like this:
| 7-day cycle | Exam countdown | |
|---|---|---|
| What the numbers mean | Day numbers, with day 1 the day you learn a topic | Steps counted back from the day before the exam |
| When you study | Days 1, 2, 3, 5 and 7 | 18, 11, 6, 3 and 1 days before the exam |
| Gaps between sessions | 1, 1, 2 and 2 days | 7, 5, 3 and 2 days, then 1 day to the exam |
| Shape | Gaps grow (expanding) | Gaps shrink (contracting) |
| Total span | One week | 18 days, ending at the exam |
| Best suited to | New material during term | A fixed exam date two to three weeks away |
Look at the gaps row. Same four digits, opposite shapes. In the 7-day cycle the gaps get longer, which memory researchers call an expanding schedule. In the countdown they get shorter, a contracting schedule. Researchers have compared those shapes head to head, with mixed results, which the research section below walks through.
Why Spaced Revision Beats Cramming
In 1885 the German psychologist Hermann Ebbinghaus published experiments in which he memorized lists of nonsense syllables and measured how quickly he forgot them. His forgetting curve showed memory dropping fastest right after learning, then leveling off. Every spaced repetition schedule, the 2357 method included, is an attempt to catch material on that curve before it slides too far.
The evidence that spacing works is about as solid as anything in educational psychology. In 2006, Nicholas Cepeda and colleagues pooled 839 assessments of distributed practice from 317 experiments in 184 articles for Psychological Bulletin. Spreading study sessions out beat massing them together, and the best gap between sessions got longer as the time until the test got longer. Keep that second finding in mind; it's the one the 2357 method has to answer to.
In 2013, John Dunlosky and four colleagues rated ten common study techniques in Psychological Science in the Public Interest. Only two earned their top rating of "high utility": practice testing and distributed practice. A 2357 timetable is distributed practice by design, and if you make each session a test, you're running both at once. Highlighting and rereading, two of the most popular study habits, were rated low.
Cramming feels more effective than it is, and Nate Kornell measured that feeling in 2009 with GRE-style vocabulary on flashcards. Studying one big stack, which spaces out each card, beat studying four small stacks one after another for 90% of participants. Yet after the first session, 72% of them believed the small stacks had worked better, a textbook illusion of competence. Spacing also beat cramming, meaning massing all the study into the last day before the test.
For 2357, that means the sessions where a topic feels shaky are the ones doing the work. A review that takes effort, because you've half forgotten the material, tends to do more for memory than one that feels smooth because you saw the notes an hour ago.
Does the 2357 Method Work? What the Research Says
Mostly, yes, if your exam is within a few weeks and every session is a test. Studies haven't tested 2357 by name, so the evidence comes from research on its parts, and three lines of it matter here. One caveat first: these studies mostly used word pairs and facts, not whole GCSE topics, so treat them as the best guide available rather than a guarantee.
Spacing matters far more than whether the gaps grow or stay equal. In 2021, Alice Latimier, Hugo Peyre and Franck Ramus pooled 29 studies in Educational Psychology Review. Spacing retrieval practice out beat massing it by a wide margin (g = 0.74 after correcting for publication bias; g is an effect size, where 0.5 is conventionally medium and 0.8 large). Growing gaps versus equal gaps made no reliable difference (g = 0.034), and shrinking gaps weren't part of that comparison. The Leitner system guide walks through the meta-analysis in more detail, including the one condition where growing gaps started to look better.
Shrinking gaps have one strong result and some contrary ones. In a 2014 Memory & Cognition paper, Carolina Küpper-Tetzel, Irina Kapler and Melody Wiseheart had 210 participants learn 28 word pairs in three sessions spread across seven days, on one of three schedules:
- Contracting: a 5-day gap, then a 1-day gap
- Equal: 3 days, then 3 days
- Expanding: 1 day, then 5 days
Participants took a final test 15 minutes, 1 day, 7 days or 35 days after their last session. At 15 minutes the three schedules tied. When the test came 1 day later, the authors report free recall 23% higher on the contracting schedule than on the other two, and at 7 days, 21% higher. At 35 days the result flipped, and the equal and expanding schedules came out 43% ahead. Equal and expanding never differed from each other.
The authors called this the first demonstration that shrinking gaps can win, and earlier studies point the other way. A 1927 study by Tsai found growing gaps did best on tests 3 and 7 days later, and Gerbier and Koenig (2012) found shrinking gaps lost to growing ones on a test two days after the last session. The 2014 experiment also used three sessions in a week, not five over 18 days. So the countdown's shape is a reasonable bet for a close exam, not a proven advantage.
The best gap depends on how long you need to remember. In 2008, Cepeda and colleagues taught more than 1,350 people a set of facts and gave them one review after a gap of up to 3.5 months. Then they tested them up to a year later. For any test date, scores rose as the gap grew, peaked, then slowly declined, and the peak moved later as the test moved further away. As a share of the time until the test, the best gap was about 20-40% for a test a week away and about 5-10% for a test a year away.
Put the three findings together and each version gets a verdict:
| Version | Gaps | What the research suggests | Verdict |
|---|---|---|---|
| 7-day cycle | 1 to 2 days, growing | For a test a week away, the best single gap was about 1.5 to 3 days (Cepeda, 2008) | A good first week with new material, but too short on its own for an exam a month or more away |
| Exam countdown | 7 days down to 1, shrinking | In one experiment, shrinking gaps won when the test came 1 or 7 days after the last session (Küpper-Tetzel, 2014); other studies disagree | A sound plan for a fixed exam date, but weak for anything you need long after the exam |
Both versions have a place, and they answer different questions. The countdown is a sensible default when you have an exam date: it guarantees five spaced sessions, the last one the day before the exam. The 7-day cycle is the better habit during term, as long as you don't treat day 7 as the end.
What to Do in Each 2357 Session: Test, Don't Reread
A 2357 timetable tells you to show up on day 5. It doesn't tell you what to do once you're there, and the default (open the notes, read them again) gets far less out of the session.
The clearest evidence comes from Jeffrey Karpicke and Henry Roediger's 2008 study in Science. Students learned 40 Swahili-English word pairs on a computer, flashcard style, under four conditions.
A week later, students who had kept retrieving every pair, including pairs they'd already gotten right, recalled about 80%. Students whose pairs were dropped from testing once they'd recalled them scored 36% and 33%. Extra rounds of studying pairs they could already recall, by contrast, did nothing for recall a week later. Every group predicted it would remember about 50%, so the students had no idea which method was helping them.
So make every 2357 session a retrieval session. A few ways to do it:
- Blurt it. Write everything you remember about the topic on a blank page, then check it against your notes and fill the gaps in a different color.
- Answer questions. Turn your notes into questions on day 1 and answer them from memory in each session. This is active recall in its simplest form.
- Do exam questions. Past paper questions are retrieval under exam conditions, which is why The Complete University Guide's countdown makes its last session a past paper.
- Explain it out loud. Teach the topic to a friend, or an empty chair, without looking at your notes. Then check what you skipped.
Whatever you choose, check your answers against the source afterward. Retrieval does the work, but a wrong answer you never correct can stick.
Here's a sample plan for one topic on the 7-day cycle:
| Day | Session | Roughly |
|---|---|---|
| 1 | Learn it: read or watch, highlight what matters, write 5 to 10 questions | 45 to 60 min |
| 2 | Blurt the topic from memory, then check and fill gaps | 15 min |
| 3 | Answer your questions without notes and flag every miss | 15 min |
| 5 | Blurt again, this time including how the ideas connect | 15 min |
| 7 | Timed exam-style questions on the topic | 20 to 30 min |
Adjust the times to the topic: dense ones need longer, and a short vocabulary list needs less.
How to Make a 2357 Revision Timetable
For a fixed exam date, build the countdown version:
- List every exam and its date. Put them all on one calendar, paper or digital.
- Mark the day before each exam as a past paper session.
- Count back from that day. Go back 2 days, then 3 more, then 5 more, then 7 more, and mark a session at each stop. You'll land on 3, 6, 11 and 18 days before the exam.
- Split big subjects. If one session can't cover a whole subject, run a separate countdown for each paper or unit.
- Fix clashes by moving a session a day. It's a small change, and studies comparing equal and growing gaps found the exact pattern mattered less than the spacing itself.
- Use the empty days for weak topics. The Complete University Guide suggests using spare days for whichever topics or subjects need them, and your past paper results will show you where those are.
Here's a worked example with three exams in June:
| Session | Days before exam | Biology (exam 8 June) | Chemistry (exam 11 June) | History (exam 15 June) |
|---|---|---|---|---|
| 1 | 18 | 21 May | 24 May | 28 May |
| 2 | 11 | 28 May | 31 May | 4 June |
| 3 | 6 | 2 June | 5 June | 9 June |
| 4 | 3 | 5 June | 8 June | 12 June |
| 5 (past paper) | 1 | 7 June | 10 June | 14 June |
Two things show up straight away. Some days carry two subjects (28 May, 5 June), which is fine. And Chemistry's fourth session falls on the day of the Biology exam, so move it to 9 June (which then holds History too) or keep it short that evening.
To turn this into a reusable spreadsheet template, put the exam date in cell B2 and use =B2-18, =B2-11, =B2-6, =B2-3 and =B2-1 for the five sessions, formatting the cells as dates. For the 7-day cycle, put the day you learned the topic in B2 and use =B2+1, =B2+2, =B2+4 and =B2+6.
If you run the 7-day cycle during term, watch the pile-up. Learn one new topic a day and, from day 7 on, every day carries that new topic plus four reviews of older ones. At 15 minutes a review, that's an hour of review a day before you've learned anything new. Plan for it, or start new topics every other day.
Where the 2357 Method Falls Short
It stops too soon for long-term memory. Four reviews in a week, or five sessions in 18 days, can carry you to an exam this month. That isn't enough for material you need next term, for a final that covers the whole year, or for a language. The further away the test, the longer the best gap: for a test about a year out, Cepeda's team put it at roughly three to four weeks.
Harry Bahrick and three members of his family, all psychologists, tested long gaps on themselves in a nine-year study published in 1993. They learned 300 foreign-language word pairs, relearned them in 13 or 26 sessions spaced 14, 28 or 56 days apart, and were tested up to five years after training ended. Thirteen sessions 56 days apart gave about the same retention as 26 sessions 14 days apart. Half the sessions did the same job, though they took twice as long on the calendar.
When day 7 ends, keep going with longer gaps: two weeks, then a month.
It treats every topic the same. A topic you nailed on day 2 gets the same four reviews as one you blanked on. Flashcard systems like the Leitner box and Anki adapt to each item and send misses back to short gaps. You can borrow the idea: if a review goes badly, treat that day as a new day 1 for that topic and restart the cycle.
A busy timetable can hide rereading. Four review sessions a day feel productive even when every one of them is a reread. If a session has no question, no blank page and no past paper in it, it's probably rereading.
2357 Method vs Leitner vs Anki: Which Schedule Fits?
The 2357 method schedules topics. Flashcard systems schedule individual facts. That difference decides which one you want.
| Method | When you review | Adapts to you? | Best for | Watch out for |
|---|---|---|---|---|
| 2357, 7-day cycle | Days 2, 3, 5 and 7, counting the learning day as day 1 | No | New topics during term | Stops at day 7 |
| 2357, exam countdown | 18, 11, 6, 3 and 1 days before the exam | No | A fixed exam two to three weeks away | Weak for anything needed after the exam |
| Leitner box | Each box on its own cycle; misses go back to Box 1 | Yes, per card | Vocabulary, formulas and definitions on paper cards | Box 1 floods if you add cards too fast |
| Anki | Per card, based on how you rate each answer | Yes, per card | Big decks you need for months or years | Reviews snowball if you skip days |
| Glasp spaced repetition calculator | 1, 3, 7, 14, 30, 60 and 120 days after you start; with an exam date, later reviews are dropped and a final one lands the day before | No | A longer expanding plan you can put in your calendar | Fixed intervals, not adaptive |
| Cramming | Everything at the last minute | No | Nothing you need to keep | Fast forgetting, and it feels like it worked |
Anki's long-standing default scheduler is based on SuperMemo's SM-2 algorithm. Since version 23.10, released on 1 November 2023, Anki has also supported FSRS, an open-source scheduler that models how likely you are to remember each card and times the next review to match.
A sensible split for most students:
- Use 2357 for whole topics and essay subjects, where the exam asks you to rebuild a topic from memory. A blurt in each session fits naturally.
- Use flashcards (Leitner or Anki) for the atomic facts inside those topics: dates, formulas, vocabulary, definitions.
- Use both for a subject like biology: a 2357 countdown for each unit, with a deck of key terms running alongside.
Running the 2357 Method on Your Highlights and Videos
Day 1 is where the material comes from, and these days that's often a mix of PDFs, web pages, Kindle chapters and lecture videos. The review sessions only work if you can quickly find what you decided mattered on day 1.
One caution first: Dunlosky's review rated highlighting low on its own. Its value here is as the capture step that feeds your retrieval sessions, not as the studying itself.
A simple setup:
- Capture on day 1. Highlight the key ideas as you read with Glasp's web highlighter, which works on web pages and PDFs and keeps every highlight in one place. For lectures and explainer videos, YouTube Summary gives you a summary and a transcript with timestamps. Reading on a Kindle? Import your Kindle highlights so they sit next to everything else.
- Turn highlights into questions. A highlight is an answer. Write the question it answers, and you have your day 2 and day 3 material.
- Use highlights as the answer key, not the study material. In each session, blurt or answer first, then open your highlights to check. Rereading them first turns a retrieval session back into a rereading session.
- Keep going after day 7. Glasp's free spaced repetition calculator takes a start date and an optional exam date and gives you a review schedule you can download as an .ics file for Google Calendar, Apple Calendar or Outlook. It doesn't follow 2357: it schedules reviews 1, 3, 7, 14, 30, 60 and 120 days after your start date. Enter the day you first learned the topic and skip its 1-, 3- and 7-day reviews, which overlap your 2357 week; the 14-, 30-, 60- and 120-day reviews carry on where the week stops.
Frequently Asked Questions
What is the 2357 method?
The 2357 method is a revision schedule based on spaced repetition. In its forward form, you learn a topic on day 1 and review it on days 2, 3, 5 and 7. A second version counts back from an exam and places sessions 18, 11, 6, 3 and 1 days before it.
Does the 2357 method actually work?
Mostly, yes, for an exam in the next few weeks. Spacing study sessions out is one of the best-supported findings in learning research. The evidence on the countdown version's shrinking gaps is mixed: a 2014 experiment found they worked best when the test came 1 or 7 days after the last session, while earlier studies found growing gaps did better. The method works best when each session is a self-test rather than a reread.
How do you make a 2357 revision timetable?
Write down your exam date, plan a past paper for the day before, then count back 2, 3, 5 and 7 days and plan a session at each point. That puts your first session 18 days before the exam. For a reusable spreadsheet template, put the exam date in one cell and subtract 18, 11, 6, 3 and 1 from it.
How many days before an exam should I start the 2357 method?
The countdown version starts 18 days before the exam. If you have less time, shorten the gaps and keep as many sessions as you can. If you have more, add longer gaps at the start (a month, then two weeks) and finish with the 2357 countdown.
What should I do in each 2357 session?
Test yourself. Blurt the topic onto a blank page, answer questions from memory or do past paper questions, then check against your notes. In a 2008 study of vocabulary learning, students who were tested on every word in every round recalled about 80% a week later, while students whose recalled words were dropped from testing managed 33-36%.
What do you do after day 7 of the 2357 method?
Keep reviewing, with longer gaps: roughly two weeks, then a month, then every few months. The 2357 method covers a week, or 18 days in the countdown version, and long-term memory needs gaps measured in weeks.
Is the 2357 method better than Anki or the Leitner system?
They do different jobs. The 2357 method schedules whole topics on fixed days, which suits essay subjects and exam countdowns. Anki and the Leitner system schedule individual flashcards and adapt to how well you know each one, which suits vocabulary, formulas and definitions. You can run both at once.
Can I use the 2357 method for GCSEs and A-Levels?
Yes. It suits an exam season with a fixed timetable: run one countdown per subject or paper, and move clashing sessions by a day. Pair it with blurting for essay subjects and flashcards for key terms.
Put Your First 2357 Cycle on the Calendar
The 2357 method is useful because it turns spacing, one of the most reliable findings in memory research, into a handful of dates. Pick the version that fits: the 7-day cycle for new material during term, the countdown for an exam date. Then make every session a test, and don't stop at day 7 if you need the material for longer than a few weeks.
Start with one topic you learned today. Highlight what matters with Glasp, write five questions, and put days 2, 3, 5 and 7 in your calendar before you close the book.
