Does counting sheep work? What the Oxford study actually found

One sheep. Two sheep. Three sheep.

It is the oldest sleep advice there is, and most people have tried it at least once. It is also one of the few pieces of folk wisdom that has been put to a proper test. In 2002, two researchers at the University of Oxford ran an experiment on people with insomnia, and the result has been quoted in every sleep article since: counting sheep does not work.

That is roughly true. But almost every retelling of the study gets the details wrong, including the part about sheep.

What the study actually tested

The paper is by Allison Harvey and Suzanna Payne, published in Behaviour Research and Therapy. It is worth being precise about it, because the version that circulates online has drifted a long way from the original.

The researchers worked with adults who met criteria for insomnia. Over several nights at home, participants were given one of three instructions for what to do once they got into bed:

Imagery distraction. Picture a situation they found interesting, engaging and pleasant, in as much detail as possible. A beach, a walk somewhere familiar, a place they liked being.

General distraction. Occupy their mind with something other than their own worries, choosing the method themselves.

No instruction. A control group, left to do whatever they normally did.

Sleep was measured using sleep diaries, so the numbers are what people reported, not what a laboratory recorded.

Two things about that list are worth noticing. The general distraction group picked their own method, so nobody in the study was told to count sheep. And the published paper does not mention sheep anywhere. The sheep came from the press coverage in January 2002, which turned “general distraction” into the most familiar example of it and gave the story its headline. The final sample was 41 people, though nearly every article since has repeated the figure of 50 that appeared in the original news reports.

None of this makes the finding wrong. It just means the study is usually described as something narrower than it was.

What they found

The imagery group fell asleep roughly twenty minutes sooner than on nights when they used no technique. That is a large difference for an intervention that costs nothing and requires no equipment.

The general distraction group and the control group did slightly worse than their own baseline. Not dramatically, but they took a little longer to fall asleep, not less.

So the honest summary is this: deliberately occupying your mind with something vivid helped. Occupying it with something vague or dull did not, and may have made the night slightly worse.

Why dull distraction backfires

The explanation the researchers offered is about capacity. Falling asleep requires the mind to let go of whatever it is chewing on. A mental task can help with that, but only if it takes up enough room.

Counting is not demanding. You can count to four hundred while simultaneously replaying an argument from work, rehearsing tomorrow’s meeting and calculating how many hours of sleep you have left if you drop off right now. The counting continues on autopilot in one corner of your attention while the worry carries on everywhere else.

A detailed imagined scene is different. To hold one you have to keep supplying material: the temperature, the sound, what is under your feet, what is in front of you. That work occupies the same attention the worry wants.

There is a second reason dull tasks fail, which the study also touched on. When the task is boring, you notice that it is not working. And noticing that it is not working is itself a thought about sleep, which is the category of thought that keeps people awake. Counting sheep quietly turns into counting how long you have been counting sheep.

What the study does not prove

This is one study from 2002 with 41 participants, measured by self-report. That is a real result, but it is not the last word, and you should be suspicious of any article that presents it as settled science. Several limits are worth stating plainly:

• Sleep diaries are not polysomnography. People with insomnia tend to underestimate how much they slept, and a technique that changes how the night feels may show up as a change in reported sleep onset even if the physiology shifted less.
• The general distraction condition was not standardised. Participants chose their own approach, so “general distraction” covers a wide range of things people happened to try.
• It tested a single night-level instruction, not a treatment. It tells you something useful about what to do in the moment. It does not tell you how to resolve insomnia that has lasted months.

The finding has held up mainly because it fits a larger body of work on what keeps insomnia going, in which pre-sleep mental activity and monitoring your own wakefulness play a central role. It is consistent, not conclusive.

What to do instead

If counting is not enough, the useful direction is a task that is more absorbing but still meaningless. Meaningful is the trap: planning tomorrow is absorbing too, and it will keep you awake for hours.

Build a scene, not a number. This is the technique that won in the study. Choose somewhere you know and like, and fill it in slowly. What the light is doing. What you can hear. What the ground feels like. When you notice you have drifted off the scene, return to it without commentary.

Or let the images be random. A newer technique, the cognitive shuffle, takes the same idea further. Instead of holding one scene, you move through a stream of unrelated, ordinary objects: a barrel, a lantern, a pebble, a window. Each one gets pictured briefly and dropped. It is harder to turn into worry than a single scene, because there is no thread to follow. We built a free tool that runs the sequence for you, with a voice so you can keep your eyes closed.

Or stop trying to fall asleep. If effort itself has become the problem, the more useful move is to lie still and gently try to stay awake. It sounds absurd, and it is a recognised technique with a clinical history. We wrote about paradoxical intention separately.

If counting sheep works for you, keep counting

Group averages are not instructions. Some people do settle with counting, and if you are one of them, the study is irrelevant to you. Nothing in it says counting is harmful. It says that, on average, it did not beat doing nothing.

The reasonable conclusion is not “counting is wrong”. It is “if counting has not worked for you after a fair try, the problem is not that you are counting badly. Try something that takes up more room.”

When the technique is not the issue

If you have been sleeping poorly three or more nights a week for longer than three months, no bedtime trick will be the answer, and searching for a better one can become part of the problem. Chronic insomnia has treatments that work and keep working. Waking up tired even after a full night is a different signal again, and worth taking seriously.

We have put together what to do about that, and how to ask for it.

Common questions

Does counting sheep work at all? On average it did not beat doing nothing in the one study that examined this kind of distraction directly. Some individuals find it helps. If you are not one of them, a more absorbing mental task is the better bet.

Did the Oxford study really test sheep? No. The published paper does not mention sheep. It compared imagining a detailed pleasant scene against general distraction that participants chose for themselves, and against no instruction. The sheep came from press coverage of the study.

Why does imagining a scene work better than counting? Because it takes more attention to sustain. Counting leaves enough spare capacity for worry to continue alongside it.

Is it bad to count backwards or count breaths instead? Counting breaths is a different thing: attention on the body rather than on a number, and it comes from a different tradition. If it settles you, it is fine. The study’s finding is about distraction being too thin to hold attention, not about numbers being forbidden.