REM sleep: how much is normal, what disrupts it and how trackers estimate it

Rapid eye movement (REM) sleep is the stage where the eyes dart under closed lids, most skeletal muscles go slack, and brain activity looks closer to waking than to deep sleep. Most vivid dreaming happens here. REM periods are short early in the night and lengthen towards morning, which is why cutting sleep short at the end of the night cuts REM disproportionately.

How much REM is normal

In that young, healthy lab sample REM was about a quarter of the night. Older adults usually get a smaller share, according to the 2004 meta-analysis. Short nights cut REM most, because the longest REM periods come last.

Alcohol and REM

The clearest everyday disruptor of REM is alcohol. Ebrahim and colleagues reviewed every study they could find on alcohol and sleep in healthy volunteers, published in 2013.

• At all doses, alcohol shortened the time to fall asleep and consolidated the first half of the night, then increased sleep disruption in the second half.
• The delay to the first REM period was the most recognisable effect and appeared at all doses.
• Total-night REM percentage fell in most studies at moderate and high doses; at low doses there was no clear trend.
• High doses also significantly reduced REM in the first part of the night.

On a tracker that often looks like a fast sleep onset, a solid early block of deep sleep, less REM and more wake towards morning. The alcohol and health page covers the wider evidence.

Other things that move REM

• Sleep length and timing. Waking early with an alarm removes the REM-rich final hours.
• Some medicines change sleep architecture, REM included. Do not change a prescribed medicine because of a sleep score; talk to the prescriber.
• Breathing. Disturbed breathing fragments sleep, REM included. Loud snoring and gasping need a sleep study, not a tracker.

How well do wearables measure REM?

REM is easier to infer than deep sleep in some respects: heart rate becomes more variable and movement is near zero. But errors on individual nights are still large.

Averages near zero hide nights that are off by 30 to 60 minutes in either direction. Software updates also change results: the same brand appears with a positive bias in one study and a negative bias in another, years apart. The most reasonable reading is that a weekly REM average from one device is informative, and a single night is not.

Is low REM a problem?

A tracker reporting 50 minutes of REM on one night is not evidence of a problem. A consistently low figure across weeks, alongside poor daytime function, is worth looking at. The first questions are the plain ones: how long are you in bed, do you drink in the evening, are you on a medicine known to affect REM, and do you snore? A clinician can arrange a sleep study if needed.

How devices infer REM

Wearables do not see eye movements. They look for the signature REM leaves on other signals: almost no movement, because muscles are slack, combined with a heart rate that is faster and more irregular than in deep sleep, at a point in the night when REM is likely. The Oura developers' own 2021 analysis showed why the extra signals matter: four-stage accuracy was 57 percent with movement alone and 79 percent when heart rate, HRV, temperature and circadian timing were added. That was the company's own data and model; independent studies of earlier hardware found lower agreement.

Newer algorithms have changed the picture for some devices. The 2024 Oura-funded study of its Gen3 ring found REM was underestimated by only 4 to 6 minutes against home polysomnography in 96 adults.

Reading REM on your own device

• Use a 7-day average. Single nights carry errors of 30 minutes or more.
• Read it with total sleep. A low REM figure on a six-hour night is mostly a short night.
• Log alcohol and bedtime. They explain more REM variation than anything else most people can change.
• Do not compare brands. One device's REM bias was positive and another's negative in the same lab on the same night.

Logging it

Peptide AI brings REM minutes in from Oura, WHOOP, Garmin, Apple Health or Health Connect through wearable integrations. With alcohol and meals logged in nutrition tracking and doses in the dose log, a pattern such as lower REM after evening drinks shows up on one screen.