The Ventilation Trials: Small Effects, Real Signal

The advice to sleep with a window cracked traces back to one small dormitory experiment. The studies since are more mixed than the advice suggests. This page walks the intervention literature honestly: what the anchor trial measured, how large the effects were, which later trials came back null, and what it means for one bedroom.

The Anchor Trial: Strøm-Tejsen 2016

Nearly every line of bedroom-air advice traces back to one 2016 paper in Indoor Air. Strøm-Tejsen and colleagues ran two field experiments in student dormitory rooms in Denmark — 14 occupants in the pilot, 16 in the second. Each slept a week per condition in balanced order, logged by wristwatch actigraphs and morning questionnaires.

In the pilot, opening a window did it: average overnight CO₂ fell from 2,585 to 660 ppm — roughly ten times the air exchange. In the second, an inaudible supply fan ran when CO₂ passed 900 ppm, taking the average from 2,395 to 835 ppm with no perceptible noise or draught, and temperatures did not differ between conditions.

What improved: actigraphy-measured sleep quality, perceived air freshness, next-day sleepiness and concentration, and a logic-task score — all significantly better in the ventilated weeks. It lacked polysomnography, clinical endpoints, and long follow-up.

What “Improved Sleep Quality” Actually Means

The phrase sounds bigger than the measurement behind it. The anchor trial — like most of the field since — assessed sleep with actigraphy, a wristwatch that infers sleep from movement, not a lab's polysomnography. Actigraphy is practical — and noisy enough to disagree with how a night felt. The sleep-tracking topic owns reading small signals honestly.

The differences were significant but modest — a few percentage points in deep sleep and awakenings, not transformed nights. The field's 2025 review is candid: when studies asked people how they slept, subjective ratings changed less consistently, with about half finding no effect. A small objective signal can be real and still sit below the threshold of feeling.

• 🔬 Movement, not brain waves — actigraphy is a proxy; polysomnography is the reference standard and most bedroom trials skipped it.
• 📏 Small margins — conditions separate by a few percentage points in group averages, not hours of extra sleep.
• 🗓️ A week per condition — long enough to detect a group difference, short enough for night-to-night variation to hide it.

The Replications and the Nulls

Does the result hold outside dormitories? The most direct test ran in 40 Danish bedrooms in winter (Fan et al., Building and Environment, 2022): an ordinary week, then a week with windows or doors flipped — 29 bedrooms changed enough to analyze. Windows open: CO₂ fell and sleep was longer and rated better. Doors open: CO₂ fell just as far and nothing improved — a door connects the bedroom to the dwelling, not to fresh air.

The trials, nulls included:

The strongest replication came in 2023: a single-blind, four-week intervention in 29 Belgian bedrooms, fan speeds covertly changed week by week (Science of The Total Environment). Where settings clearly separated CO₂, higher ventilation meant more deep sleep and fewer awakenings; across the 23 bedrooms with a clean high-versus-low contrast, deep sleep was significantly shorter at the low setting. Cognition never differed — and only some rooms produced contrasts clean enough to analyze.

Then the nulls, which are why this page exists. A 2024 summer study in Shanghai (50 bedrooms) ran the same logic in heat: open windows lowered CO₂ but brought in heat, PM2.5, and noise — REM sleep got shorter — and open doors did nothing. The 2025 review of 17 papers agrees, conditionally: ventilation affects sleep on average, but the effect depends on season, city, and what the incoming air carries — a case for revisiting standards, not a settled rule.

Why Effects Are Small

None of this makes the positive trials wrong — sleep is noisy, ventilation a modest lever. Why they look small:

• 🎯 The rooms at issue are the bad ones — benefits should be largest where the baseline is worst; an airy bedroom has little to win.
• 📊 Small samples, varied sleepers — 14 to 36 people can reliably detect only sizeable shifts; a modest effect may miss significance in any study.
• 🌡️ Many inputs, noisy measures — temperature, noise, light, and schedule move the same actigraphy numbers, and week-long windows blur the rest.
• 🕰️ It predates the nulls — a smaller earlier dormitory crossover, discussed in the 2016 paper, found no significant effects.

The Dose Question

If ventilation helps, how much does it take? No personal threshold is validated — but a working ladder sits on real measurements. 36 healthy adults slept two nights at each of three levels — 750, 1,000, and 1,300 ppm average overnight CO₂, balanced order (Building and Environment, 2024). Sleep efficiency fell 1.3–1.8% at the higher levels; time awake rose about 5 and 8 minutes.

• 🪜 Where the line sits — the 2025 review found no disturbance below 1,000 ppm and proposed keeping bedrooms at or under 800 ppm as a precaution, given sensor error; a target for standards, not a switch. The measurement page owns what a reading can and cannot tell you.
• 📟 Real rooms run far above it — a 2026 study of 20 damp-flagged Wellington bedrooms found CO₂ indicating inadequate ventilation; an earlier survey saw peaks up to 4,000 ppm.
• 🧭 The practical stance — lower is plausible-better and cheap where outdoor air is clean and quiet; the how-to page owns the trade-offs.

What Would Settle It

Three upgrades would move this from promising to settled:

• 🧪 Bigger randomized trials — controlled studies run 14 to 36 people; the Belgian trial needed most of its 29 rooms to find analyzable contrasts.
• 🛌 Lab-grade measurement — polysomnography on a subsample, so effects are read from sleep architecture, not movement.
• 🌍 Many climates and seasons — Danish winter and Shanghai summer behaved differently; single-site results should not stretch across geographies.
• 🔬 The pollution side — the door-opening null suggests CO₂ is a marker, not the mechanism; trials varying pollutants independently would isolate what ventilation does.

Until then, the case rests on plausibility, small positive trials, and a near-zero price. The rest of the room is on the bedroom-engineering page.

Questions, Answered Briefly

• ❓ Did any trial show a large sleep improvement? No. Positive results are modest group-level shifts — deep-sleep minutes and fewer awakenings, not extra hours in bed.
• ❓ Why do some trials find nothing — is CO₂ the problem? Cutting the number is not the same as fixing the air: door-opening lowered CO₂ without changing what the sleeper breathed, and the review notes no study has shown pure CO₂ below 5,000 ppm disturbing sleep.
• ❓ What about next-day focus? Two of eleven studies that measured it found gains — the 2016 anchor and the 2022 window arm. Possible, not promised.
• ❓ So should I sleep with the window open? If the room is stuffy and outdoor air is clean and quiet, it is a cheap, plausible-better default; if it is smoky, hot, or loud, the calculus flips.

Related Topics