REM Cycle Calculator

REM cycle estimator

REM is back-loaded: the last cycles of the night hold most of it.

Time asleep
8h in bed
7h 45m
Complete cycles
plus a 15m partial cycle
5
Estimated REM
~21% of sleep
1h 36m
Cost of cutting one cycle
Removed from the end of the night
-31m REM

Estimates from population averages for REM distribution. Only a sleep study measures your actual REM.

REM is not spread evenly. That single fact changes how you should think about a short night, and it is the reason the calculator above weights the later cycles far more heavily than the early ones.

How REM builds across the night

How REM builds across the night
CycleRoughlyDominated by
15% REMDeep sleep (N3)
210-15% REMDeep sleep, less of it
3~20% REMBalanced
4~28% REMREM and light sleep
5-635-42% REMREM, almost no deep sleep

The practical consequence: sleeping six hours instead of seven and a half does not cost you 20% of everything. It costs you almost no deep sleep and a third of your REM.

What that means for a short night

If you have to lose sleep, lose it at the start rather than the end. Going to bed 90 minutes later and keeping your alarm preserves the REM-heavy cycles. Keeping your bedtime and setting the alarm 90 minutes earlier removes them. Both give you the same total; they are not the same night.

The same logic explains REM rebound. After a few short nights, the pressure for REM builds and the next full night front-loads it. That is also why the dreams after a period of bad sleep, or after stopping alcohol, feel unusually vivid.

What actually suppresses REM

  • Alcohol. Suppresses REM early and fragments the rebound later. Two drinks is enough to show up.
  • Short nights. By far the largest effect, for the reason above.
  • SSRIs and SNRIs. Well-documented REM suppression. Not a reason to stop a prescription, but worth knowing when reading your tracker.
  • Cannabis. Reduces REM while used, with rebound on withdrawal.

A note on wearable REM numbers

Consumer devices infer stages from movement and heart rate variability. They are reasonable at total sleep time and poor at distinguishing REM from light sleep, with validation studies showing wide error margins against polysomnography. Watch the trend across weeks; do not react to a single night's REM figure.

Related: how the 90-minute cycle works, and the sleep cycle calculator for choosing how many cycles to aim at.