On this page
Spreading revision out beats cramming. The harder question is how far apart the sessions should be. A 2008 study in *Psychological Science* answered it: it depends on how long you need to remember.
What did Cepeda et al. (2008) study?
The paper is Cepeda, Vul, Rohrer, Wixted and Pashler (2008), "Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention", Psychological Science, volume 19, pages 1095-1102. It mapped the spacing effect over weeks and months, not the minutes used in most earlier lab work.
The results cover 1,354 people from the lab's online research panel, living in many countries, who completed all three sessions; their mean age was 34 (range 18-72) and 72% were female, according to the full text hosted by the first author.
The design had three sessions:
Learn 32 obscure but true trivia facts until each was recalled correctly once.
After a gap of 0 to 105 days, review: answer every fact twice more, seeing the correct answer after each try.
After a further test delay of 7, 35, 70 or 350 days, take a recall test and then a five-option multiple-choice test, with no feedback.
Each person was randomly assigned to one of 26 gap-and-delay combinations, and total study time was the same for everyone. Only the timing changed.
What is the optimal gap between study sessions?
For every test delay, scores rose as the gap grew, peaked, then slowly fell. The paper estimated the best recall gap by fitting smooth curves to the data:
Test in 7 days: best gap about 3 days (43% of the delay).
Test in 35 days: best gap about 8 days (23%).
Test in 70 days: best gap about 12 days (17%).
Test in 350 days: best gap about 27 days (8%).
So the best gap gets longer as the exam gets further away, but it shrinks as a fraction of that time. The abstract puts it as roughly 20-40% of a one-week delay falling to about 5-10% of a one-year delay.
The effects were large. Compared with reviewing straight away, the best gap gave a 64% increase in final recall (d = 1.1), as the paper reports. At the 70-day test delay, recall improved by 111%.
What does the study not show?
It used trivia, not problem-solving. Recalling a one-word answer is not solving a rotational mechanics question.
The learners were adults online, mostly women with an average age of 34, not Indian teenagers in coaching or school.
There was only one review. It says nothing directly about a third or fourth pass.
People dropped out more at longer delays. The authors found completers and non-completers looked similar on initial knowledge, age and background.
Later work supports the main pattern. A companion lab study by the same group, Cepeda et al. (2009) in Experimental Psychology, used foreign vocabulary, facts and object names with test delays up to 6 months and found an optimal gap improved recall by up to 150%. And Kim and colleagues (2019) in Behavior Research Methods analysed workplace-training data from 10,514 people and again found the optimal spacing grew with the retention interval. For how spacing ranks against other study techniques, see our explainer on the 2013 Dunlosky review of ten study techniques.
How should a JEE or NEET student use this?
What follows is our inference, not a finding of the study. The authors' own advice: decide how long you want to remember something, then choose the gap.
A worked example. Say you finish Electrostatics today, and you need it for a test in 10 weeks (about 70 days). The study's best recall gap for a 70-day delay was about 12 days, so plan your review roughly 10-14 days from now, not the next morning. If the exam were a year away instead, the 350-day result points to a single review about 4 weeks out instead.
A rule of thumb:
Test in about a week: review after 1-3 days.
Test in about a month: review after about a week.
Test in about two to three months: review after about 10-14 days.
Test in about a year: review after about 3-4 weeks.
Make each review a self-test (close the book, write what you remember, then check); the study's review was a test with feedback, not rereading. Our guide to why cramming the night before fails explains the short-term trap, and this six-week board revision calendar with 1-3-7-14 day gaps turns the idea into dates.
If working out what to review on a given day is the hard part, Fermi's AI tutor works out what you already know and builds a plan for the session, so you sit down to a clear next step.



