Ask someone who has pulled an all-nighter how they feel, and they will often say they are tired but "still fine". Laboratory studies tell a less reassuring story. Sleep loss reliably slows simple reaction time and increases the number of long lapses in attention, and people are often poor at noticing how impaired they have become. That makes sleep one of the most important variables behind a reaction-time score — and one of the easiest to misread.
The bottom line
Short-term sleep deprivation degrades simple attention and reaction time more than it degrades complex reasoning. The effect is not just a slower average; it shows up as more lapses, more variability, and a higher chance of a very slow response. The standard laboratory task for this is the psychomotor vigilance test (PVT), and the best summaries of the evidence come from meta-analyses and controlled sleep-restriction experiments rather than from a single night's self-experiment.
What the PVT measures
The psychomotor vigilance test is a simple reaction-time task: a signal appears at unpredictable intervals, and the participant responds as quickly as possible. It is deliberately boring, because boredom is the point — the task stresses sustained attention rather than problem-solving. Researchers use it because it is sensitive to sleep loss and because it can be administered repeatedly under controlled conditions.
The key outcome is not only the mean reaction time. PVT analyses often focus on lapses — responses that are unusually slow or missed — and on response-speed metrics. Those measures are more sensitive to sleep deprivation than a simple average, which can be pulled around by a few extreme values.
What the evidence shows
A meta-analysis of short-term total sleep deprivation found significant effects across cognitive domains, with the largest effect for lapses in simple attention and the smallest, nonsignificant effect for reasoning accuracy. In other words, sleep loss makes it harder to stay vigilant and respond quickly; it does not necessarily make complex reasoning collapse in the same way. The review covered 147 cognitive tests from 70 articles, and its authors noted that time awake was a significant predictor of between-study variability for accuracy measures.
A separate study designed to maximize the PVT's sensitivity found that metrics related to lapses and psychomotor speed had high effect sizes under both acute total sleep deprivation and five nights of partial sleep restriction. Mean and median reaction time — the metrics most people intuitively focus on — were less sensitive and more influenced by extreme values. The practical implication is important: two people can have similar average times while one has many more lapses, and the lapses may be the more meaningful sign of sleep loss.
Acute versus chronic sleep loss
One bad night is not the same as two weeks of short nights. A controlled study of chronic sleep restriction found that attention and reaction-time performance degraded cumulatively across days, even when participants spent only part of the night awake. The deficit accumulated, and participants' subjective sleepiness did not track the objective decline perfectly. That is why "I feel fine" is weak evidence that your reaction time is unaffected.
Acute total deprivation is more dramatic, but chronic partial restriction is more common in ordinary life: late nights, early starts, and a few hours of debt repeated across a week. The research suggests that this pattern can produce measurable impairment even when it does not feel like an emergency.
How large is the effect?
It depends on the task, the amount of sleep lost, and the person. A frequently cited comparison is that performance after prolonged wakefulness can resemble performance at a legal alcohol limit, and the original comparison is worth reading rather than quoting as a slogan. The broader point from the meta-analytic and PVT literature is that sleep loss produces a real, measurable slowing and an increase in lapses — not a subtle effect that only a laboratory can detect.
An important caveat: effect sizes from controlled studies are averages. Some people are more resilient to a given amount of sleep loss; others are more vulnerable. Averages describe groups, not your individual response on a particular morning.
Why your browser test may not show the full picture
The reaction test on this site is a short, five-attempt task. It can show a slower average when you are tired, but it is not the PVT. It has fewer trials, a different signal schedule, and a shorter duration, so it has less opportunity to catch occasional lapses. A normal-looking five-attempt average does not prove that your attention is intact; a slow average does not prove that sleep is the cause.
Device latency is another confound. If your setup adds tens of milliseconds, a sleep-related change of a few milliseconds is partly hidden. Compare like with like: same device, same time of day, several rounds, and a note of how much you slept. The direction of a consistent change across days is more informative than any single score.
How to test the relationship for yourself on the reaction time test
- Record sleep, not just the score. Note bedtime, wake time, and roughly how rested you felt.
- Test at a consistent time. Reaction time varies across the day; comparing a morning score with a late-night score mixes circadian effects with sleep debt.
- Run several rounds and use the median. A five-attempt average is a sample; a median across rounds is more stable.
- Do not change multiple things at once. If you also start a new caffeine routine or a new mouse, you will not know what moved the number.
- Look for lapses, not only the average. If your slowest attempts are much slower than usual while your fastest ones are unchanged, that pattern is worth noticing.
- Do not use the result to self-diagnose a sleep disorder. Persistent sleep problems, excessive daytime sleepiness, or concerns about your health belong with a qualified professional.
Reading the pattern, not just the average
Suppose your five attempts are 210, 215, 220, 225, and 410 ms. The average is 256 ms, but the story is in the last attempt: four responses were close to your normal range and one was a long lapse. A sleep-deprived state often looks like that — not a uniform slowdown of every response, but a higher chance of an occasional very slow one. If you only glance at the average, you may miss the pattern that matters most.
The opposite pattern is also informative. If every attempt shifts upward by a similar amount — say 20–30 ms slower across all five — that is more consistent with a general slowing than with a single attentional lapse. Neither pattern proves sleep is the cause, but they suggest different things to investigate: a higher floor points toward a state or device change, while isolated long lapses point toward attention and vigilance.
Across days, look at the distribution rather than one round. A single slow morning can be noise; three mornings in a row with a wider spread and more very slow attempts is a pattern worth taking seriously. The same logic applies to your fastest attempts: if your best times are stable but your worst times are much worse, the deficit is in consistency, not raw speed. If you want a rough personal check, run one normal round and look at the spread between your best and worst attempt; a sudden widening of that spread after a short night is more informative than a small shift in the average.
What to do with the result
If your scores are consistently slower after short sleep, the useful response is not to grind more rounds or to chase a personal best with caffeine. Protect sleep, test again under better conditions, and treat the difference as a signal about your state rather than a verdict about your ability. Sleep is not a performance hack you can skip; it is part of the measurement conditions. The conservative interpretation is also the useful one: a consistently slower score after short sleep is a reason to protect sleep, not a reason to train harder on a tired brain.
For the controllable side of reaction time — practice, warm-up, device setup, and caffeine — see the Improve Reaction Time guide and the caffeine evidence review. For how the test itself is scored, see Methodology.
Limits of this evidence
PVT studies are laboratory experiments with controlled sleep schedules, standardized tasks, and relatively small samples compared with population surveys. They tell us that sleep loss impairs vigilance and reaction time; they do not tell us exactly how your score should change after a specific night, and they are not a clinical assessment. Treat this guide as orientation for interpreting your own measurements, not as medical advice.
For how other state factors compare with sleep loss, see the guides index.
References
- Lim, J., & Dinges, D. F. (2010). A meta-analysis of the impact of short-term sleep deprivation on cognitive variables. Psychological Bulletin, 136(3), 375–389. PubMed record.
- Basner, M., & Dinges, D. F. (2011). Maximizing sensitivity of the psychomotor vigilance test (PVT) to sleep loss. Sleep, 34(5), 581–591. PubMed record.
- Van Dongen, H. P. A., Maislin, G., Mullington, J. M., & Dinges, D. F. (2003). The cumulative cost of additional wakefulness. Sleep, 26(2), 117–126. PubMed record.
- Dawson, D., & Reid, K. (1997). Fatigue, alcohol and performance impairment. Nature, 388, 235. PubMed record.