The Forgetting Curve Explained: Why You Forget What You Study (and How to Beat It)

You sat through the lecture. You highlighted the textbook. You even nodded along, certain it was going in. Two weeks later you open your notes for the exam and half of it feels like you’re reading it for the first time. That is not a lack of intelligence or effort – it is a well-documented, measurable pattern called the forgetting curve, and once you understand the shape of it, you can work with it instead of fighting it. For the broader study system this guide sits inside, see our full guide to studying effectively at Australian universities.

The forgetting curve is a model, first plotted by German psychologist Hermann Ebbinghaus in 1885, showing that memory for newly learned information fades fastest in the hours immediately after learning, then slows down over time. Left un-reviewed, most of what you study drops away within days. Reviewed at the right intervals through spaced repetition and active recall, that same information can be pushed into durable long-term memory – which is exactly why this guide covers both the science and the study system built on top of it.

What is the forgetting curve?

The forgetting curve describes how quickly newly learned information is lost from memory when there is no attempt to retain it. It is not a straight line. Forgetting happens fastest in the first hour and first day after learning, then the rate of loss slows down the longer the memory survives – so the curve looks steep at the start and gradually flattens out.

The concept comes from one specific person and one specific study: Hermann Ebbinghaus, who in 1885 published Uber das Gedachtnis (“On Memory”), the first attempt to measure memory loss using controlled, repeatable experiments rather than guesswork or introspection. Everything people now call “the forgetting curve” traces back to that book.

Three things make the forgetting curve useful rather than just a piece of psychology trivia:

• It explains why cramming the night before an exam produces knowledge that vanishes within days of sitting the test.
• It gives you a rough timetable for when a memory is about to be lost – which is the entire basis of spaced repetition, the study method covered later in this guide.
• It has been independently retested more than a century later, using modern statistical methods, and the basic shape held up (see the 2015 replication study below).

Who discovered the forgetting curve? Hermann Ebbinghaus's 1885 experiment

Hermann Ebbinghaus was a German psychologist working in Berlin in the 1880s, at a time when memory was considered too subjective to study scientifically. He disagreed, and set out to prove it by becoming his own experimental subject – memorising and testing himself, alone, over several years.

To stop his existing knowledge and associations from skewing the results, Ebbinghaus invented what he called nonsense syllables: three-letter, consonant-vowel-consonant combinations that resembled words but meant nothing, such as “WID”, “ZOF”, “DAX” and “RIY”. He built a pool of roughly 2,300 of these and drew lists of them at random for each test, specifically to strip out meaning, familiarity and emotional association – the things that normally make some information easier to remember than other information.

His method, known as the savings method, worked like this:

1. Learn a list of nonsense syllables to a strict criterion: two consecutive, completely error-free recitations.
2. Record exactly how many repetitions (or how much time) that took.
3. Let a set amount of time pass, with no further review, ranging from 20 minutes to as long as a month.
4. Relearn the same list to the same criterion, and record how many repetitions that took this time.
5. Calculate the “savings”: the percentage reduction in effort between the first learning session and the relearning session.

A simple worked example makes the savings method concrete: if it took 10 repetitions to learn a list the first time, and only 5 repetitions to relearn it a day later, that is a 50% saving. A high savings score means most of the original memory was still intact, even if Ebbinghaus could not consciously recall the list before he started relearning it. A low savings score means most of the memory had genuinely been lost.

Plotting the savings percentage against the time delay produced the curve. It dropped sharply within the first hour, continued falling (more slowly) across the first day, and then flattened out considerably over the following days and weeks – the classic “steep then shallow” shape people now mean when they say “the forgetting curve”.

Ebbinghaus also worked out an equation to describe that curve mathematically. His published savings formula (Ebbinghaus, 1913 edition, p.77) is:

b = 100k / ((log t)^c + k)

Where b is the percentage saved (his measure of retained memory), t is the time in minutes since roughly a minute after the end of learning, and k (1.84) and c (1.25) are constants he fitted to his own data. This is worth knowing because most study-advice articles skip straight to a simplified textbook formula (covered in the next section) without mentioning that Ebbinghaus’s own published equation was more complex, and was derived from his actual results rather than assumed in advance.

Is the forgetting curve real? What the 2015 replication study found

Yes – and this is one of the more interesting parts of the story, because Ebbinghaus’s 1885 result was not just accepted on faith for 130 years. In 2015, memory researchers Jaap M.J. Murre and Joeri Dros published a formal replication in the peer-reviewed journal PLOS ONE (Murre & Dros, 2015, Replication and Analysis of Ebbinghaus’ Forgetting Curve, PLOS ONE 10(7):e0120644).

Rather than nonsense syllables, Murre and Dros used Dutch-German word pairs, and one of the researchers (Dros) put in roughly 70 hours across the study, learning and relearning lists at six intervals: 20 minutes, 1 hour, 9 hours, 1 day, 2 days and 31 days after initial learning – a wider spread than Ebbinghaus originally tested. The result: the shape of the new curve was, in the authors’ own words, “similar to Ebbinghaus’ original data”.

One genuinely new finding is worth flagging because it rarely makes it into shorter summaries of the study: the replicated curve was not perfectly smooth. The researchers found a small jump upward in retention starting around the 24-hour mark, suggesting sleep, or some other consolidation process that happens overnight, may briefly slow the rate of forgetting – a detail Ebbinghaus’s original, more limited testing schedule was not positioned to catch.

Two conclusions follow from the replication. First, the basic finding – fastest forgetting soon after learning, slowing over time – is not a one-off artefact of one obsessive 19th-century psychologist’s data. Second, and less widely reported, forgetting is not a perfectly clean mathematical curve in practice; real memory decay has bumps in it, shaped by things like sleep, that a single smooth equation cannot fully capture.

What is the forgetting curve formula?

Most study-advice websites quote a simplified version of Ebbinghaus’s work rather than his actual published equation. The version usually shown is:

R = e^(-t/S)

Where R is retrievability (how likely you are to successfully recall the memory right now), t is the time elapsed since learning, and S is the relative “stability” of the memory – a rough stand-in for how well it was originally learned and how many times it has been reviewed. A higher S means a flatter, slower-decaying curve; a lower S means a steeper drop.

This exponential formula is genuinely useful as a mental model and is the basis of the retention estimate in the calculator further down this page. But it comes with an important, rarely mentioned caveat: it is a simplification, not what Ebbinghaus himself published, and a 2011 study by Lee Averell and Andrew Heathcote (Journal of Mathematical Psychology 55(1):25-35) directly compared exponential decay against an alternative, power-law decay model, and found the exponential model does not fit real forgetting data as well as the power-law alternative does – particularly over longer time periods. In plain terms: forgetting slows down more, later on, than the neat exponential curve predicts.

The chart below plots the shape both models broadly agree on – a steep early drop, then a long, shallow tail – using an indicative starting retention of 100% at the moment learning ends, assuming no review at all.

Read that chart as a warning, not a life sentence. It shows what happens with zero review. The entire second half of this guide is about how spaced repetition and active recall interrupt that curve on purpose, at the exact points where forgetting is about to accelerate.

How fast do you actually forget? What the numbers really show

You have probably seen a specific claim somewhere online: that you forget 50%, 70% or even 90% of new material within a day or a week. Treat any single, precise percentage you see quoted without a source, including on some corporate training blogs, with real scepticism – Ebbinghaus’s own results varied by material, and the 2015 replication (covered above) found real forgetting has bumps and irregularities a single tidy percentage cannot capture.

What can be said with confidence, because it holds across both the original 1885 data and the 2015 replication, is the pattern:

• The steepest drop in memory happens in the first 20 minutes to few hours after learning, well before you have even gone to bed on it.
• A second, smaller drop typically happens over the first one to two days.
• After that, the rate of loss slows considerably – what survives the first day or two tends to be far more durable than what does not.
• Reviewing material before it is fully forgotten, rather than after, is dramatically more efficient than relearning it from scratch once it is gone – this is precisely what the savings method above was measuring.

Does the forgetting curve apply to everyone the same way?

No, and this is one of the more genuinely debated parts of memory science. Three separate lines of evidence complicate the idea of one universal curve that applies equally to all memories and all people.

Not all forgetting follows one mathematical shape. Memory researcher John T. Wixted, in a widely cited 2004 review (Annual Review of Psychology 55:235-269), concluded that no single decay model explains all the ways human memory actually behaves – forgetting is shaped by emotional weight, how well something was originally encoded, sleep, and interference from other things learned around the same time, not time alone.

Everyday, personally meaningful memories do not decay the same way as memorised lists. In an unusually long-running study, psychologist Marigold Linton recorded two personal-event memories every day for six years and tested her own recall of them over time. Her forgetting turned out to be close to a straight line – she lost roughly 5-6% of items per year – rather than the sharp-then-flat curve Ebbinghaus found with nonsense syllables. The likely explanation is that meaningful, personally relevant memories are encoded differently from arbitrary lists memorised for a lab experiment, which is a useful reminder that deliberately meaningful study material (making real connections, not just repetition) may resist the forgetting curve better than rote memorisation does.

How well you originally learned something changes the shape of your personal curve. Material learned to a shallow, first-pass level decays fast. Material that was over-learned – reviewed past the point where you could first recite it – decays much more slowly, which is a large part of why spaced repetition (below) deliberately revisits material rather than only ever teaching it once.

How to beat the forgetting curve with spaced repetition

Spaced repetition is the practice of reviewing material at gradually increasing intervals – short at first, then longer – timed to land just before you would otherwise have forgotten it. Instead of one long cram session, you get several short, well-timed touches. It is the single most consistently evidence-backed countermeasure to the forgetting curve, and it has a surprisingly long, well-documented history.

1. 1932 – C.A. Mace, “Psychology of Study”. The first person to formally propose spacing out revision sessions rather than massing them, suggesting a schedule of roughly “one day, two days, four days, eight days, and so on”.
2. 1939 – H.F. Spitzer. Tested spaced review on close to 3,600 sixth-grade students in Iowa, one of the earliest large-scale field tests showing spaced review outperformed massed study in a real classroom.
3. 1967 – Paul Pimsleur, “graduated-interval recall”. Published exact expanding intervals for language learning: 5 seconds, 25 seconds, 2 minutes, 10 minutes, 1 hour, 5 hours, 1 day, 5 days, 25 days, 4 months, then 2 years – still the conceptual basis of the Pimsleur language courses.
4. 1973 – Sebastian Leitner, the Leitner system. A physical flashcard box with five compartments of increasing size (1cm, 2cm, 5cm, 8cm, 14cm). Cards you get right move to the next, less-frequently-reviewed compartment; cards you get wrong drop back to the first, most-frequently-reviewed one – a simple, self-adjusting spacing system anyone can build with an index card box.
5. 1990s onward – algorithmic spaced repetition software (SRS). Piotr Wozniak’s SuperMemo was the first widely used software to calculate personalised review intervals automatically, based on how well you performed on each review. Its ideas underpin modern apps like Anki, which since version 23.10 uses a newer scheduling algorithm called FSRS (Free Spaced Repetition Scheduler), and are echoed in the review mechanics of language apps like Duolingo.

You do not need an app to use this. The Leitner box system above works with nothing but index cards and a shoebox, and it captures the same core principle every algorithmic tool is built on: review things you find easy less often, and things you find hard more often, with the gap between reviews growing every time you get something right.

Research backs the general approach beyond any one tool. A 2007 review by Pashler, Rohrer, Cepeda and Carpenter (Psychonomic Bulletin & Review) found that, for most material, the ideal gap between study sessions is roughly 10-20% of how long you need to remember it for – so for an exam six weeks away, review sessions spaced roughly four to eight days apart tend to outperform both cramming everything at once and spacing sessions too far apart. A 2021 meta-analysis by Latimier, Peyre and Ramus (Educational Psychology Review) went further, describing spaced retrieval practice as a “strong recommendation” for both teachers and students based on the accumulated evidence.

What is active recall, and why does it work better than re-reading?

Active recall is the practice of testing yourself on material – trying to pull the answer out of your own memory – instead of passively re-reading notes or a textbook. It is spaced repetition’s essential partner: spacing tells you when to review, active recall determines how that review should actually work, and the evidence is clear that the “how” matters just as much as the “when”.

The key study here is Roediger and Karpicke (2006, Psychological Science 17(3):249-255), often referred to as the “testing effect” study. Students read a short science passage, then either reread it repeatedly or tested themselves on it through free recall, without being graded. One week later, on a final test, the group that had tested themselves – even though it felt harder and less “smooth” at the time – substantially outperformed the group that had simply reread the passage. Re-reading felt more comfortable in the moment; it produced weaker long-term memory.

This is sometimes called “desirable difficulty” in memory research: the mental effort of trying to retrieve an answer, even when you get it wrong, strengthens the memory trace more than passively seeing the correct answer again does. Every time you successfully retrieve a memory, the neural pathway involved is reinforced – similar to how walking the same trail through a forest repeatedly keeps the path clear, while an unused trail grows over. Turning your own notes into recall-friendly material also makes a real difference here – see our guide to the best note-taking methods for university lectures.

Practical ways to build active recall into study sessions, roughly in order of how quickly they are to set up:

• Closed-book recall. Close the textbook or notes and write down everything you remember about a topic before checking what you missed.
• Flashcards (physical or digital). One question per card, answer on the back – the format practically forces retrieval rather than recognition.
• Practice questions and past exam papers. As close as you can get to simulating the real retrieval conditions you will face in the exam itself.
• The “teach it to someone else” test. If you can explain a concept out loud, from memory, in your own words, without glancing at notes, you have genuinely retrieved and reconstructed it rather than just recognised it.
• Blurting. Set a timer, write down everything you can remember about a topic in one uninterrupted burst, then check your notes for gaps.

A practical study schedule based on the forgetting curve

Turning the theory above into an actual weekly routine matters more than memorising the history. A schedule that works with the forgetting curve, rather than against it, generally follows this shape for any new topic covered in a lecture, tutorial or reading:

Three adjustments that matter for international students studying in Australia specifically:

• Most Australian university teaching periods run 12-13 teaching weeks per semester, so a 12-topic subject naturally gives you roughly one new topic per week – build the review schedule above into your weekly routine from week one rather than starting it in swot vac (revision week), when there is no longer time to space anything out properly.
• If your degree is heavy on tutorial-based or case-study assessment rather than pure recall exams, weight your active recall sessions toward practice questions and applying concepts to new scenarios, not just reciting definitions – the forgetting curve applies to skills and applied reasoning too, not only facts.
• Swot vac (study/exam preparation week) is for consolidation, not first-time learning. If a topic is still being learned for the first time in swot vac, the spacing benefits above are already mostly lost – it becomes cramming by necessity, which still works better than nothing, but produces memory that decays fast once the exam is over.

For more on structuring study time across a full semester, including how to balance this with assignments and part-time work, see our guide to time management for international students in Australia, and for how to pick between studying with a group or solo for a given topic, our guide to managing assignments, exams and group work.

Where else the forgetting curve shows up beyond the classroom

The forgetting curve is not just an exam-study concept – it shows up anywhere someone is expected to retain information they were only told once, which turns out to be a surprisingly wide range of situations.

Medical information. A study published in the Journal of the Royal Society of Medicine (Kessels, 2003, 96(5):219-222) found that patients forget between 40% and 80% of the medical information given to them by a doctor almost immediately after the appointment ends – and of what they do remember, a meaningful share is remembered incorrectly. It is a large part of why clinics increasingly follow up with written instructions and spaced text-message reminders rather than relying on a single verbal explanation during a consultation.

Eyewitness memory. A 2008 meta-analysis by Deffenbacher, Bornstein, McGorty and Penrod (Journal of Experimental Psychology: Applied 14(2):139-150), covering 53 separate studies, found a consistent relationship between how long a witness waits before making an identification and how much their memory for a face has degraded – a real-world, high-stakes example of the same decay pattern Ebbinghaus first plotted with nonsense syllables, which is part of why police interviews are generally conducted as soon as possible after an incident.

Workplace training and onboarding. New employees who receive a single induction session and no follow-up typically retain only a fraction of it within weeks, which is why well-run onboarding programs now build in scheduled refreshers rather than treating induction as a one-off event – directly applying spaced repetition outside an academic setting.

Skills training, including surgery. A 2023 pilot study published in the Journal of Neurosurgery found that spaced repetition improved neurosurgery residents’ retention of a procedural skill compared with standard, unspaced training – a reminder that the forgetting curve applies to physical, hands-on skills, not only facts and definitions.

Common myths about the forgetting curve

Retention calculator: estimate how much you will remember

This calculator uses the simplified exponential model covered earlier (R = e^(-t/S)) to give a rough, illustrative estimate of how much of something you would still retain after a given number of days with no review, based on how well you originally learned it. Treat the output as a teaching tool that shows the shape of the curve, not a precise personal prediction – as covered above, real forgetting varies by person and by material, and the actual rate for any one memory can run faster or slower than this simplified formula suggests.

As a general anchor for the “moderate” setting above (one properly focused study session, no further review): this model estimates retention falls to roughly 50% by around day 2, and to around 10% by about day 7 – which is exactly the window the practical study schedule earlier in this guide is built to interrupt with a review before that drop happens.

Frequently asked questions