How often to review lecture notes: a spacing schedule

How often to review lecture notes depends on how long the material has to last. For an exam in a week, review every one to three days. For an exam in a month, review about once a week. For a course you sit an exam on in a year, review a few days after the lecture and then about once a month.

Spaced review is the practice of repeating material at intervals that grow with the distance to the test, instead of repeating it all in one evening. The interval is the whole method. The review itself is ordinary work.

What the spacing research actually shows

Cepeda and colleagues published a meta-analysis in Psychological Bulletin in 2006. They collected 839 assessments of distributed practice from 317 experiments reported in 184 articles. Two of their findings matter when you plan a semester. Spacing beat massing consistently: episodes spread over time retained better than the same episodes run back to back. And the best gap was not a constant. The interval that produced maximal retention grew as the retention interval grew.

That second result is the one people ignore. "Review it three times" is not a plan, because three reviews inside one week and three reviews across three months are different treatments with different results.

The experiments behind that meta-analysis used verbal material: words, passages, facts. That is close to what a lecture gives you, and it is why the result carries over to study notes rather than staying in a laboratory. The studies are about timing, and timing is the one part of a study plan you control before you know whether the method works.

How far apart the reviews should be

Cepeda and colleagues ran a second study, published in Psychological Science in 2008, to put numbers on the gap. More than 1,350 people learned a set of facts. They returned for a review after a gap of up to 3.5 months, and they were tested after a further delay of up to one year. Retention first rose and then fell as the gap lengthened, so there was a peak. The peak moved with the test delay: the longer the delay, the longer the optimal gap.

Measured as a proportion of the test delay, the optimal gap shrank. The authors report it at about 20 to 40 percent of a one-week delay and about 5 to 10 percent of a one-year delay.

You can use that proportion without the original data. Take the time left before the test and place the gap between a tenth and a third of it. A test in seven days points to a gap of roughly one to three days. A test in one year points to gaps of roughly three to five weeks. That arithmetic is mine, not theirs; the study reports proportions.

How often to review lecture notes after each lecture

The first review belongs within 24 to 48 hours, while the lecture can still be reconstructed. Do not review on the night of the lecture if you can avoid it. You still hold the material in working memory that evening, so your own notes look obvious. The next day you find out what they actually say.

Three steps are enough:

• Within 48 hours, close the notes, say out loud what the lecture covered, and only then check.
• After three to seven days, answer the questions you wrote, not the answers you wrote beside them.
• Then widen the gap. If the answers are complete, double the interval. If they are not, repeat at the same interval.

A review calendar for one recorded lecture

A recording makes the calendar practical. You do not have to choose what to review by flipping through pages. You search the transcript for a term and land on the minute where it was explained.

For a lecture recorded on a Monday, with an exam 21 days later:

• Tuesday: open the transcript, write six questions, answer them from memory.
• Friday: answer the same six without the transcript open.
• The following Tuesday, day 8: answer them again and add three new questions.
• Day 15: answer all nine in one sitting, timed, without notes.
• Day 20: final pass over the three answers you still get wrong.

Five contacts with the material in three weeks, and only one of them is a reread.

Lectures on the same topic do not need separate calendars. The third lecture on one subject should reuse two of the six questions from the second, because the gap between the two lectures is already an interval you did not have to schedule. Courses run in series; study plans usually do not, and that is a large part of the waste.

How often to review lecture notes before an exam

Work backwards from the exam date rather than forwards from the lecture. Write the exam day on the calendar first. Put the last review two days before it, not on the night before: that night is for sleep and for the two facts you keep mistaking.

Then move backwards in gaps of about a tenth to a third of the remaining time. For an exam in three weeks the working gaps are seven days, then three days, then one day. For an exam in two days, one review is all that fits, and it should be a test rather than a reading.

How many reviews one lecture needs

There is no fixed count. What is fixed is the total number of contacts, and a three-hour recording does not have to be one unit. Split it into three spans of about an hour and give each span its own six questions. Each span then keeps its own calendar, and a missed week damages one third of the lecture instead of all of it.

Once the questions exist, a review costs about ten minutes. Six to ten questions, answered aloud, checked against the transcript, is a complete review. Reading the same lecture again costs three hours and produces less.

If you have twenty lectures and three weeks, do not schedule five reviews for every one. Pick the lectures whose transcript you could not reconstruct without help, give those the full calendar, and accept a single pass over the rest. A review you actually do beats a schedule that collapses in week two.

What counts as a review and what does not

Roediger and Karpicke (2006, Psychological Science) tested the difference directly. Students read prose passages and either took one or three immediate free-recall tests, with no feedback, or restudied the passages the same number of times. A final test came five minutes, two days or one week later. After five minutes, repeated studying produced better recall. On the delayed tests, prior testing produced substantially greater retention than studying, even though the students who restudied were more confident.

That order is the trap. Rereading wins immediately and loses later, and you feel competent the whole way through. A review that feels smooth is evidence about the moment, not about the exam.

The tests in that study were free recall with no feedback, so the students never saw the right answers. Testing helped retention anyway. You do not need a perfect answer key to run a review. You need the attempt first and the check second, in that order.

A schedule built from the spacing ratio

The proportions come from Cepeda et al. (2008). The counts are arithmetic from those proportions, not measured optima. Treat the table as a starting grid and adjust it to what your answers show.

Review schedules that fail

• One long session the night before. This is massing, the condition that spacing beat.
• Highlighting. It produces a review-shaped activity with no retrieval in it.
• Three reviews in four days and then nothing for a month.
• Reviewing only the parts you already answer correctly.
• Watching the recording again at 1.5x and calling it a review.

How to keep the calendar across a semester

Every lecture gets one line: the date, six questions, the next review date. When a review is due, it is due, whether or not you have finished the new lecture. A missed review moves forward by one interval, never to the end of the semester. Ten lectures with five reviews each is fifty contacts, and that is a semester's work, not an evening's.

When two reviews a week is the wrong answer

• You cannot state the main point of the lecture in one sentence, so there is nothing to space yet.
• The questions were copied from the slides and only ask for definitions.
• You answer by glancing at the transcript halfway through the sentence.
• The exam is in two days, and what you need is one honest test, not a calendar.
• The material is a procedure, and you have never worked it end to end.
• Your transcript is machine-made and its key terms are wrong.