Best temperature for sleep: what research says about bedroom heat, cold and AC at night

People ask for the best temperature for sleep and expect a number to set on the thermostat or the air conditioner. This article went to the research looking for one and did not find it. Three studies are used here: a review of how the thermal environment affects sleep, a US survey of nighttime temperature and insufficient sleep, and a Korean study of nighttime temperature and sleeping-pill prescriptions.

It is informational material, not medical advice. Sleep that stays bad for weeks is a question for a doctor, not for a thermostat.

Best temperature for sleep: the single number is missing

None of the three sources names an ideal bedroom temperature. Figures such as 18 °C or "16 to 19 °C" are widely quoted as ideal, and they do not appear in these papers. What the sources do contain is more useful and less tidy, and it comes down to four things: what you sleep under, how humid the heat is, how cold the room gets, and what the air conditioner is doing at which hour.

The main source is a 2012 review by Okamoto-Mizuno and Mizuno. Its starting point is that heat and cold both tend to produce more time awake and less REM and slow-wave sleep, but the size of the effect depends heavily on bedding and clothing (Okamoto-Mizuno & Mizuno, 2012). For how heat, light and pressure compare as weather inputs to sleep, see weather and sleep. This article is about the narrower question of the room itself.

Why bedding changes the answer

The review gives one reference point that sounds like a recommendation and is not. For a semi-nude person, the thermoneutral air temperature is 29 °C, and air above or below it brings more wakefulness and less REM and slow-wave sleep. That is a laboratory condition without blanket or clothes, so it is not a setting for your air conditioner.

In real life people sleep under covers, and the picture changes. In studies with clothing and/or bedding, no significant difference in sleep was found across air temperatures of 13 to 23 °C and of 3 to 17 °C. In older adults, actigraphy showed no difference in sleep quality between 9 and 20 °C. The review's reading is that, with covers, cold in this range does not change sleep stages.

There is also a second temperature, the one under the blanket. The review says the bed climate is usually around 32 to 34 °C and 40 to 60% relative humidity when sleep is normal. The room air is only part of what your body is dealing with. Bedding sits between you and the room, which is why the same room air does not mean the same bed climate.

Humid heat adds to the thermal load

With bedding in play, heat increases wakefulness and reduces slow-wave sleep and REM, and the review states that humid heat adds to the thermal load further. It describes an experiment that compared a steady 26 °C and 60% humidity with a switch to 32 °C and 80% in one part of the night. At 32 °C and 80%, slow-wave sleep was affected most, while REM was kept.

For heat and humidity during the day, see heat, humidity and fatigue. The humidity of a room is something you can estimate, and the site has a calculator for indoor humidity from outdoor weather. It was built for winter, so read it as an estimate for heated rooms.

Sleeping with AC on all night: what the one relevant experiment says

Searches such as "sleeping with AC on all night" and "best temperature for sleep with AC" ask something the research answers only in part. None of the three papers tested air-conditioner settings, and none compared an all-night run with a partial one in ordinary bedrooms.

What the review does report is a habit. In Japan, most people run the air conditioner for only a few hours after falling asleep, because they think it is bad for health and to save energy. From the 26 °C versus 32 °C experiment above, the authors conclude that if use is limited, the air conditioner is better used at the start of sleep. If it runs in the second half of the night, they say it is important to dry off sweat and change clothing, to avoid "chilling effects".

That is the authors' conclusion from one laboratory experiment with humid heat. It is not a rule that all-night cooling is harmful or safe.

A frequent follow-up is whether air conditioning gives you a cold or a sore throat. These studies say nothing about that. Whether a chill can lead to a cold is a separate question, covered in does getting cold give you a cold. For babies and small children, ask a pediatrician: this data does not cover them.

Best temperature for sleep in winter: the cold you do not feel

The review's most unexpected point is about cold. With bedding, cold exposure did not change sleep stages. It did affect the cardiac autonomic response during sleep, and it did so without changes in sleep stages or in how people said they felt. When air temperature fell from 17 to 3 °C, heart rate variability measured during sleep changed.

On that basis the authors argue that in real life the impact of cold may be greater than that of heat, and say further studies are needed. They also write: "It is extremely important to note that Ta in winter should be maintained at a level higher than 10°C." This is a statement in a review, and it is quoted as that: not a measured threshold and not a thermostat target. The authors add that the difficulty is that sleep itself is not disturbed in the cold, so the effect can occur without any sensation. They also note that bedroom temperature in winter in Japanese suburbs can fall to 3 °C.

So an honest answer to the best temperature for sleep in winter is narrower than a number: a cold room can feel fine under a blanket and still matter, and the evidence on how much is thin.

Outdoor night temperature and sleep

Two large studies looked at the temperature outside, not in the bedroom.

A US survey of 765,000 respondents between 2002 and 2011 matched answers to nighttime temperature. Warmer nights increased self-reported nights of insufficient sleep. The effect was largest in summer, among lower-income and elderly respondents (Obradovich et al., 2017). The abstract also contains climate projections, and this article does not use them.

A Korean study of 711,079 adults found that about 6% (42,858) had been prescribed hypnotics. The relationship between nighttime temperature, taken as the district's mean highest temperature between 23:00 and 07:00, and the dose was non-linear and U-shaped. Higher doses were seen in the categories of 0 °C and below and 0 to 5 °C, and also in 20 to 25 °C and 25 °C and above, compared with a reference of 10 to 15 °C (Min, Lee & Min, 2021).

Both are observational. Both use outdoor temperature, not the temperature of any bedroom. One measures sleep by self-report, the other by prescriptions, which is not the same as how well someone slept. They show an association and not a cause. The 10 to 15 °C reference is a statistical baseline for outdoor temperature, not a bedroom recommendation.

What nobody knows yet

• Which air temperature is best under a given blanket. The review reports no significant difference across wide ranges, and no paper here picks a best value.
• Whether running an air conditioner all night is better or worse than a few hours. The one experiment on timing was in humid heat.
• How much the unfelt effect of cold on the heart matters for health. The authors call for more studies.
• Whether a US survey and Korean prescription data say anything about your own room. They concern populations and outdoor air.
• Anything about infants and children.

What you can check in your own data

The MeteoHealth app lines up your own sleep quality with outdoor weather data, such as its comfort index and the pressure change, and looks for a pattern in your history. It does not measure the temperature in your bedroom and does not know whether an air conditioner was on. It does not give norms and does not diagnose. If your sleep gets worse on hot nights, the app can show whether that lines up with outdoor heat in your entries. What happened under your blanket is something you can only track yourself.