attraction.lab

Does a Sleep Divorce Work? The One-Third Figure Counts Adults, Not Couples

A third of Americans sleep apart, says the survey everyone quotes. It asked 2,005 adults, not couples, and nobody has measured what it does to a relationship.

Research-backed writing on attraction, dating and relationships. Drafted by AI against a named bibliography, published only when every claim links to its source.
31 min read
Editorial illustration for: Does a Sleep Divorce Work? The One-Third Figure Counts Adults, Not Couples

Before the article, the one piece of advice here that has evidence behind it and is not about your relationship. If the reason one of you wants to move rooms is snoring — especially snoring with gasping, pauses in breathing, or daytime sleepiness — those are the cardinal symptoms of obstructive sleep apnoea, and they are a reason to see a doctor rather than to rearrange the bedrooms. In the one controlled study that measured both, treating the breathing disorder improved the patient's marital satisfaction. A spare room solves the noise and leaves the condition untreated in the person who has it.

With that said. “Sleep divorce” is a phrase with no research behind it and one statistic in front of it: more than a third of Americans sleep in a separate room from their partner. That figure comes from a single two-page PDF published by the American Academy of Sleep Medicine, and reading it changes the number in three ways at once. The survey asked 2,005 American adults — not couples, not people who live with a partner. It offered a tick-list of six things people do “to accommodate a bed partner,” and the famous third is two of those boxes added together. And the only box that describes an arrangement rather than an event, “sleep in another room consistently,” was ticked by 15%.

Underneath the surveys there is a real literature, and it is small, old and says the opposite of what the trend pieces say. In the only studies that measured the same people sleeping both ways, people slept objectively better alone — and reported being less satisfied with it.

What follows is fifteen sections: where the number comes from, why the men and the women in the same survey describe different bedrooms, the four conflicting prevalence estimates, the 1969 laboratory study and the 2020 one, the study nobody has run, what is actually established about sleep and conflict, the one place causal evidence exists, the single population-level finding about sleeping apart, and what the first page of search results cites. Our editorial policy explains why every empirical claim here carries a linked source.

#1

The figure everyone quotes, and the question that produced it

Almost every article on sleep divorce opens with the same statistic: more than a third of Americans sleep in a separate room from their partner. It comes from one place, the American Academy of Sleep Medicine, and the AASM publishes the underlying table as a two-page PDF. Reading it changes the number.

The methodology paragraph is four sentences long and worth quoting as written: “The American Academy of Sleep Medicine commissioned an online survey of 2,005 adults in the U.S. The overall margin of error fell within +/- 2 percentage points with a confidence interval of 95 percent. Fieldwork took place between March 24-29, 2023. Atomik Research is an independent market research agency.”

The question was: “Do you ever do any of the following to adjust your sleep routine to accommodate a bed partner?” It was a list to tick, not a question about where you sleep. The options and the counts:

  • Use earplugs — 303 (15%)
  • Use an eye mask — 358 (18%)
  • Sleep in another room on occasion — 391 (20%)
  • Sleep in another room consistently — 309 (15%)
  • Go to sleep at an earlier or later time than desired — 660 (33%)
  • Use a silent alarm — 325 (16%)
  • None of these — 840 (42%)

So the famous third is 20% plus 15%, added together from a tick-list. And the base of those percentages is 2,005 adults in the United States, not 2,005 couples, not 2,005 people who live with a partner. Nothing in the published table restricts the sample to people who have a bed partner at all; the question invites anyone to answer, and 42% ticked nothing.

That is the same move this site has taken apart in the numbers people quote about modern dating: a real survey, honestly published, whose denominator is replaced somewhere between the PDF and the headline. The AASM did not say a third of couples sleep apart. A market research panel of American adults, asked whether they ever do any of six things, produced two ticked boxes that sum to 35%.

#2

Twenty of those thirty-five points are people who did it once

The table separates two answers that the coverage merges, and the distinction is the whole question.

“Sleep in another room on occasion” is 391 people, 20%. On occasion includes the night someone had flu, the night a child was sick, the night one of them fell asleep on the sofa. It is not a sleeping arrangement. It is an event that has happened at least once.

“Sleep in another room consistently” is 309 people, 15%. That is the only line in the survey that describes an arrangement, and it is a little over one in seven American adults — not one in three, and still measured on a base that includes people with no partner.

Watch what happens to the distinction downstream. The page that ranks third for this query, Sleep Foundation's explainer, carries an AASM “Trusted Source” badge and writes that one survey found “over one-third of people regularly sleep in a different room from their romantic partner.” The word is regularly. The AASM's own regular category is 15%. The 35% only exists if you add in the people who did it once, and once is the opposite of regularly.

This is not a dispute about rounding. It is the difference between a minority arrangement and a majority-adjacent trend, and the second version is the one that gets written about. The same structure turns up whenever a number is passed along rather than looked up — the three-day rule, the 3-6-9 rule and the rest of the numbered folklore survive on exactly this kind of relay, and so does the rebound-relationship literature, where the genre's confident claim and the one study that measured it point opposite ways.

If you want the honest sentence for the top of an article about sleeping apart, the AASM data support this one: in a 2023 online panel of 2,005 American adults, 15% said they consistently sleep in another room to accommodate a bed partner, and another 20% said they do so occasionally.

#3

Men and women in the same survey describe different bedrooms

The AASM table breaks the results down by gender, and the two halves do not match.

The sample is split exactly: 997 men and 997 women. Among the men, 244 (24%) sleep in another room on occasion and 209 (21%) do so consistently — 453 people, 45%. Among the women, 147 (15%) and 100 (10%) — 247 people, 25%. A twenty-point gap.

The careful caveat first: these men and women are not each other's partners, nobody's bed partner is necessarily in the sample, and not every respondent's partner is of the other sex. So this is not literally two people disagreeing about one bedroom. But a sleeping arrangement is a household fact, and in a nationally fielded sample balanced to 997 and 997, reports of the same class of household fact should not come out nearly twice as high from one half as from the other.

And the gap is not specific to the bedroom item. Look down the column:

  • Earplugs: men 214 (21%), women 89 (9%).
  • Eye mask: men 234 (23%), women 123 (12%).
  • Silent alarm: men 246 (25%), women 79 (8%).
  • Go to sleep earlier or later than desired: men 371 (37%), women 285 (29%).
  • None of these: men 340 (34%), women 494 (50%).

On every single item, men report more accommodating than women. Half of the women ticked nothing at all. Either American men are systematically more considerate sleepers than American women, in every domain measured, or the question is picking up something about who says yes to a tick-list — what counts as doing something “to accommodate a bed partner,” what counts as another room, how far back memory is allowed to reach.

That is a familiar pattern rather than a surprising one. The most famous finding about sex differences in the psychology of jealousy turned out to depend on whether people were made to choose between two options or allowed to rate both. Self-report instruments measure the question as much as the person, which is also the central problem with the five love languages. It is a reason to treat a survey's internal consistency as evidence about the survey.

#4

The age mix decides the headline, and nobody mentions it

The second breakdown in the AASM table is by age, and it does two things at once: it shows who sleeps apart, and it shows that the aggregate figure depends on who was asked.

The cells, which sum exactly to 2,005: 18-24, n = 313. 25-34, n = 456. 35-44, n = 541. 45-54, n = 367. 55-64, n = 176. 65+, n = 152.

Now the answers. “None of these” — the people who do nothing at all to accommodate a bed partner — runs 42%, 36%, 31%, 46%, 60%, 67% across those six groups. Sleeping in another room on occasion peaks at 35-44 (145 people, 27%) and falls to 13% among the over-65s. Sleeping in another room consistently runs 13%, 19%, 18%, 16%, 7%, 9%.

So the behaviour is concentrated in the middle of the age range, and the two oldest groups barely do any of it. Which means the headline percentage is a function of how many people of each age are in the panel — and this panel is heavily weighted towards the groups that adjust most. The 35-44 cell alone is 541 people, 27% of the whole sample, the largest of the six, and it is the group with the highest rate on almost every item. The two oldest cells together are 328 people, 16%.

The survey was quota-balanced on sex to the exact person — 997 and 997 — and was not balanced the same way on age, and the published table reports no weighting. Whatever else the sample is, it is a panel of people who answer online surveys, skewed towards the middle-aged.

There is a second thing hiding in the same column, and it matters for the way the subject is usually written. A trend is supposed to be visible in the young. Here the people most likely to report sleeping in another room on occasion are in their late thirties and early forties — the ages when small children and shift work and different alarm times are most common, which are the explanations nobody needs a trend for. The age pattern in the next survey down points the other way entirely.

#5

The other surveys give other numbers, and one reverses the age pattern

If the question is how many couples sleep apart, there is more than one answer in circulation, and the pages that rank for this query each pick one.

The Better Sleep Council, whose 2017 survey Cleveland Clinic links, reports that 9% — almost one in ten — say they sleep in separate bedrooms. That is a quarter of the AASM headline. Its age breakdown runs in the opposite direction to the AASM's: the over-55s are the most likely to sleep in separate bedrooms at 16%, against 7% of 35-to-54s and 3% of 18-to-34s. The AASM panel's consistent sleepers-apart peaked in the twenties and thirties at 18-19% and fell to 7-9% in the oldest groups.

So: two surveys of American adults, both quoted by first-page results, disagree about the size of the phenomenon by a factor of four and about its direction across the life course. Neither page acknowledges the other. The Better Sleep Council page publishes no sample size, no field dates and no method, and the Council is the consumer-education arm of the mattress industry — which is not a reason to discard its numbers, but is a reason to want the method it does not print.

Sleep Foundation's own on-site survey supplies a third set: 53% of people who tried sleeping apart said their sleep quality improved, those who stuck with it reported sleeping 37 minutes more a night, over a quarter eventually went back to sharing a bed, and 40% of those said their sleep improved after they went back. Those numbers appear with no published sample size or method either, on a page that carries an inline mattress recommendation with a discount code.

The pattern is one this site keeps running into: a question gets answered by whoever is selling the adjacent product, and the figures that result are built on whoever happened to respond. It is the same defect as the recovery-time numbers in how long it takes to get over a breakup, where three of the circulating figures trace to one descriptive statistic about a sample, and as the survival rates in what happens to couples after infidelity, where the celebrated percentages are computed on subgroups of five and fourteen couples.

#6

What sharing a bed does to sleep was measured in a laboratory in 1969

Underneath the surveys there is a real experimental literature, and it is small and old and says something the trend pieces do not.

In 1969, Lawrence Monroe published a study in Psychophysiology with a title that tells you exactly what he did: “Transient changes in EEG sleep patterns of married good sleepers: the effects of altering sleeping arrangement.” Twenty-eight married good sleepers — fourteen couples — slept three consecutive nights in the laboratory under each of two conditions: sleeping with their spouse, and sleeping alone. Continuous EEG and eye-movement recordings were taken across a seven-hour bed period.

The result, in his own abstract: a significant increase in Stage 4 sleep and a significant decrease in REM sleep in the sleep-alone condition. Stage 4 is the deepest slow-wave sleep. By the usual objective markers of sleep depth, people slept more deeply when their partner was not there.

Two other things from the same paper, both reported and both ignored since. The transient changes in these good sleepers were not accompanied by the usual correlates of poor sleep — this was not insomnia, it was a different architecture. And there were sex differences within the couples: the women showed significantly more Stage 4 sleep, more total sleep and fewer awakenings than the men, which is the reverse of the picture the AASM tick-list produces when it asks people to describe themselves.

Monroe's own conclusion was methodological rather than romantic. He argued that the usual home sleeping arrangement is a subject-selection variable that sleep and dream researchers should be controlling for. Fifty-seven years later it is still the cleanest within-person comparison anyone has published of the same people sleeping both ways, and it is not what either side of the sleep divorce argument quotes — because what it says, on its own, is that the bed partner costs you deep sleep. The next section is why that is not the end of it.

#7

The part of the same literature that nobody quotes

Wendy Troxel's 2010 review in Psychosomatic Medicine is the one piece of actual research that a first-page result links to on this topic, and it contains a paragraph that inverts the trend.

Summarising the objective studies, Troxel writes that in research measuring sleep by polysomnography or actigraphy on co-sleeping nights versus sleeping-alone nights, participants generally sleep better when sleeping alone. She cites Monroe for the architecture result above, and then adds the finding Monroe's abstract leaves out: “Despite these changes in sleep architecture, participants reported being less satisfied with their sleep when sleeping alone.”

The same inversion shows up in the only other study of its kind. Frances Pankhurst and James Horne published two related actigraphy studies in Sleep in 1994. In the first, 46 pairs of bed partners aged 23 to 67 were monitored for eight nights. Men made significantly more discrete movements during sleep than women. Between 5 and 6% of all 30-second sleep epochs contained a movement, and about a third of those movements were shared by both partners within the same epoch — highest in the younger couples. Female bed partners reported being disturbed by their partner more often than male partners did. The second study compared the presence and absence of a bed partner, and Troxel's summary of it is that there were more movements during sleep when couples slept together — and that these couples also generally reported better subjective sleep quality when sleeping with their partners than when sleeping alone.

Troxel's reading of the divergence is the sentence worth carrying out of this whole literature: the findings “suggest that the psychological need for closeness and security, particularly at night, trumps the equally important need for good quality sleep.”

So the measured position is almost exactly the opposite of the popular one. The popular version says sleeping apart feels like a loss but is better for your sleep. The measured version says sleeping apart tends to be better for your sleep and feels worse, and that people have consistently chosen the worse sleep. That is a real trade-off, and it is a trade-off about preference, not a defect to be fixed by an article.

#8

The pro-sharing case rests on twelve couples in a German sleep lab

The other half of the argument — that sharing a bed is actively good for you — has one study behind it, and it is worth knowing its size.

Henning Drews and colleagues published it in Frontiers in Psychiatry in 2020. Twelve heterosexual couples, target age 18 to 29, all of whom had shared a bed with the same partner most nights for at least three months, with shift workers, pregnancy, depression, addiction and sleep disorders screened out. Each couple spent four nights in the sleep laboratory across two weekends, undergoing individual and simultaneous dual polysomnography, with the order of conditions counterbalanced.

The results, compared against sleeping individually: about 10% more REM sleep when co-sleeping, less fragmented REM (p = 0.008), longer undisturbed REM fragments (p = 0.0006) and more limb movements (p = 0.007). No other sleep stage changed significantly. Sleep architectures were more synchronised between partners during co-sleep, and that synchronisation was positively associated with the depth of the relationship. The effect on REM was largest in the individuals with the least social support.

Note first that this agrees with Monroe: both found more REM when sharing a bed, and both found more movement. Fifty-one years apart, with different equipment, the two studies point the same way on the measurements they share. That is more than most of this literature can say.

Note second what the evidence base is. Drews and colleagues write in their own introduction that only two polysomnographic studies exist comparing co-sleeping and individual sleep in healthy couples — and one of the two is their own pilot. Twelve couples, aged 18 to 29, four nights, in a laboratory. Whatever “REM sleep is better when you share a bed” means, it means it about twenty-four young German adults who already shared a bed and were wired up for the weekend.

There is a companion paper, Drews and Drews in 2021, which compared 69 married against 69 never-married individuals matched from the Sleep Heart Health Study and found more REM in the married group. That is a study about being in a couple, not about which room you sleep in, and it cannot distinguish between the two.

Further illustration for: Does a Sleep Divorce Work? The One-Third Figure Counts Adults, Not Couples
#9

Nobody has run the study the question needs

Here is the simplest way to see the size of the hole. Search PubMed — the index of the biomedical literature — for the phrase “sleep divorce.” It returns one record.

That record is Monica Andersen, Tathiana Alvarenga, Leandro Fonseca and Sergio Tufik, writing in the Journal of Health Psychology in 2025: “Interactions between co-sleeping, sleep quality, and relationship health: The benefits and challenges of shared sleep environments.” It is a narrative review, not a study. It describes sleep divorce as “a potential strategy” and closes by saying what future research should do: focus on personalised interventions, on the impact of sleep disorders on co-sleeping, and on “the long-term effects of different sleeping arrangements on relationship satisfaction and overall well-being.”

That sentence is the field saying, in 2025, in the only indexed paper that uses the term, that the long-term effects of different sleeping arrangements on relationship satisfaction have not been established. Which means nobody has done the obvious study: take couples who share a bed, have some of them sleep apart, and follow both groups. No randomisation. No before-and-after with a comparison group. Not even a cohort followed from the decision forwards.

The wider literature confirms the shape of the gap rather than filling it. Amie Gordon, Belinda Carrillo and Christopher Barnes reviewed sleep and social relationships in healthy populations for Sleep Medicine Reviews in 2021, identifying over 200 relevant articles, and concluded that “the majority of work is correlational, limiting conclusions about the directionality of these effects.”

This is a shape worth recognising, because it recurs. A term coined outside research gets treated as though a research literature stands behind it — the same thing that happened with emophilia, where eight papers and one author carry a construct the wellness press treats as established. And sometimes the question itself has simply never been asked, as with whether friends-to-lovers relationships last longer, where the only direct comparisons sit unpublished in two university repositories.

#10

What has been measured is sleep quality, and the link is real

One link in the sleep divorce argument is genuinely well supported, and it deserves saying clearly before the caveats resume: sleeping badly damages how two people treat each other the next day.

Amie Gordon and Serena Chen tested it two ways. In a 14-day daily-experience study, people reported more conflict in their relationships after poor nights of sleep. Then they brought couples into the laboratory: one partner's poor sleep was associated with a lower ratio of positive to negative affect — both self-reported and observed by coders — and with decreased empathic accuracy for both partners during a conflict conversation. Conflict resolution happened most when both partners were well rested.

The crucial detail is in their controls. The effects were not explained by stress, anxiety, depression, low relationship satisfaction, or by the partner being the source of the poor sleep. Bad sleep hurt the next day's conversation whether or not the person beside you caused it.

Three more studies in the same direction. Nicole Maranges and James McNulty followed 68 newlywed couples for seven days and found spouses more satisfied on days after longer sleep, with sleep buffering husbands' — not wives' — global satisfaction against negative specific evaluations. Brant Hasler and Troxel measured 29 co-sleeping couples for a week with diaries and actigraphy and found the links running both ways and differently by sex: for men, better sleep efficiency predicted less negative partner interaction the next day; for women, less negative interaction during the day predicted better sleep that night. And a 2023 paper in the Journal of Social and Personal Relationships, three studies totalling 695 people including a quasi-experiment, found poorer sleep quality predicting worse perceived relationship quality, with increased anger accounting for the association.

Put that together and the argument for sleeping apart has a real first half and a missing second. If moving rooms genuinely improves both people's sleep, this literature says the relationship should benefit. What nobody has shown is the if: that for a given couple, separate rooms produce better sleep rather than a different set of disturbances, and that the benefit survives the cost of the nights apart. The conflict research also points at a cheaper target than the bedroom — how two people repair things afterwards, which is where the research on reconnecting after a fight and the habits of couples who last both land.

#11

The one place where something like causal evidence exists: treating the snorer

There is a version of this question that does have a controlled study behind it, and it is not about bedrooms. It is about the thing that drives most people out of the bed.

In 2001, Tracy McFadyen, Colin Espie, Nigel McArdle, Neil Douglas and Heather Engleman published a prospective, parallel-group trial in the European Respiratory Journal. Patient-and-partner couples were taken from the top of a CPAP waiting list (n = 44, treated) and the bottom (n = 25, conservative management) and rated marital satisfaction with the ENRICH questionnaire, health and functional status, and sleepiness, with both groups' baselines taken while on conservative therapy. Baseline variables did not differ.

At follow-up, all seven summary psychosocial scores were statistically better in the CPAP-treated patients. The effect on marital satisfaction was moderate, around 0.5 of a standard deviation; the effects on health, functional status and sleepiness were very large, above 1.0 SD. Scores in the conservatively treated group deteriorated. Objective CPAP use correlated modestly with improvement across every psychosocial domain.

Two honest qualifications. Allocation was by position on a waiting list, not by randomisation, so this is a controlled prospective comparison rather than a randomised trial. And the outcome is the patient's marital satisfaction, measured in couples where one partner had sleep apnoea. Troxel's review separately describes a study by Beninati and colleagues in which co-sleeping apnoea patients and their spouses showed fewer objectively measured arousals after CPAP; we have not read that paper and report it as Troxel describes it.

Still, this is the only place in the whole subject where an intervention was applied, a comparison group existed, and marital satisfaction moved. The intervention was not a second bedroom. It was treating the breathing disorder.

Which is why the referral belongs here as well as at the top. Loud snoring, gasping or pauses in breathing during sleep, and daytime sleepiness are the cardinal symptoms of obstructive sleep apnoea, a condition with real cardiovascular consequences. A couple that solves it by moving one person down the hall has solved the noise and left the condition untreated in the person who has it.

#12

The one finding that points the other way, and it is about loneliness

There is a single population-level study that measured an outcome of sleeping in separate bedrooms, and it is not on the first page of any search result for this topic.

Chi Chiao, Wen-Hsu Lin, Yi-Hua Chen and Chin-Chun Yi published it in BMC Geriatrics in 2021, using data from 1,645 older parents in northern Taiwan. Loneliness was measured with the De Jong Gierveld short scale, which separates emotional loneliness from social loneliness. Among the married, those who slept in separate bedrooms were more likely than bed-sharing couples to report both emotional and social loneliness, and the association with emotional loneliness held after adjusting for sleep problems and, the authors state, independently of relationship satisfaction. Satisfying spousal relationships reduced the size of the association in the social domain but did not remove it.

The limits are substantial and go in both directions. This is cross-sectional: one measurement, no time order, and the obvious alternative reading is that people who already feel distant in a marriage are the ones who end up in the second bedroom. It is one region of one country. The participants are older parents, a long way from the thirty-somethings who fill the AASM panel. And sleeping apart in a long marriage may be a marker of illness, caregiving or disability as much as of choice.

It is still the only estimate anyone has of what sleeping apart goes with, at population scale, on an outcome that is about the relationship rather than about sleep architecture. And the outcome it picked is a specific one. Not satisfaction, not divorce — loneliness inside a marriage, which is the thing the Sleep Foundation's own survey gestures at when it reports that for more than a third of the people who went back to sharing a bed, what brought them back was missing each other.

Two related pieces on this site go through what else predicts that feeling: why people stay in relationships that are not working, where the investment model turns out to explain more than satisfaction does, and what longitudinal data says about thinking it might be over, where the thought itself turns out to be unstable across a single year. And the problem of which clock you measure by — the one that decides the cohabitation finding in does living together before marriage lead to divorce — applies here too, because “since when” is the question nobody asked these couples.

#13

Nights apart have been measured once, and attachment changed the result

A separate bedroom is not a business trip. But the one well-measured instance of a couple spending nights apart is a work separation, and what it found is worth knowing before anyone treats the arrangement as neutral.

Lisa Diamond, Angela Hicks and Kimberly Otter-Henderson studied 42 couples over 21 days, timed so the window contained a naturally occurring four-to-seven-day separation, collecting daily diaries and salivary cortisol. From pre-separation to separation, and again from separation to reunion, there were significant changes in the quality of partners' interactions, in positive and negative affect, in subjective stress, in physical symptoms, in cortisol — and in sleeping problems, which rose during the separation and fell at the reunion.

The moderators are the interesting part. The effects were more pronounced in the partner who stayed home, in partners high in attachment anxiety, and in partners who had less contact during the separation. And they were not moderated by relationship length, by relationship satisfaction, by how often the couple underwent separations, or by whether there were children at home.

Read that list carefully, because it is unusual. The things people assume would matter — how long you have been together, how happy you are, how used to it you are — did not. The thing that did was an individual difference in how one person's attachment system handles distance. Troxel's review makes the same point from the other end: she argues the effects of co-sleeping are likely to be moderated by partners' relational styles rather than being a fact about beds.

That is as close as this literature comes to a practical steer, and it is a steer about who you are rather than what the research recommends. Our attachment style test scores the two dimensions separately and says on its own page that it is an educational tool and not a diagnosis; what the two sides of it mean in practice is covered in supporting a partner with anxious attachment and supporting an avoidant partner, and the question of whether any of it is fixed is in can attachment styles change. The separate literature on what physical distance does to how two people see each other — including what happens when the distance ends — is in the distance effect on attraction.

#14

What the first page of results actually cites

We fetched the nine pages that ranked for “sleep divorce” on 17 November 2026 and counted their citations rather than reading their headlines. Two served nothing useful to an automated request; the rest are below. The count is of outbound links to journals, DOIs, PubMed and PMC.

Four research citations in total, across all nine pages. They are:

  • Cleveland Clinic, ranked first, links a 2021 Sleep Medicine Reviews meta-analysis of randomised trials showing that improving sleep quality improves mental health — a paper about sleep interventions, with nothing about couples in it — plus a 2009 Journal of Sleep Research paper on how 36 hours of sleep deprivation affects bargaining and trust games, plus the Better Sleep Council survey.
  • Sleep Foundation links a 2017 Psychoneuroendocrinology paper on sleep and inflammatory responses to marital conflict, a 2023 Journal of Social and Personal Relationships paper on sleep quality and anger, and the AASM release.
  • Psychology Today's piece, titled “Can a Sleep Divorce Save Your Marriage?”, is written by Elise Buie, a family-law attorney, and its single research link is a 2009 paper on the bidirectional relationship between major depression and marital disruption. Sleeping arrangements do not appear in it.
  • Oura's blog links Troxel's 2010 review — the one source on the entire page that is actually about couples' sleeping arrangements, and the one whose central paragraph, quoted above, contradicts the article it is supporting.

Simply Psychology, Verywell Mind, Marriage.com and Love What Matters cite no research at all.

And one citation deserves its own line. Sleep Foundation's claim that “one research study found that sleeping with a romantic partner was associated with both better sleep quality and better mental health” carries an Oxford University Press trusted-source badge. The link goes to a one-page conference abstract — Fuentes, Kennedy, Killgore and Wills, number 0010 in the SLEEP 2022 abstract supplement, page A4 — which as far as the indexes show was never published as a full paper.

Not one source cited anywhere on that page compared couples sleeping apart with couples sharing a bed. This is the same audit that produced the finding in do dating apps actually work, where one page of about twenty-seven results cited a study and got it wrong in two ways in a single sentence. Here, nobody got it wrong. Nobody looked.

#15

What is supported, what is not, and what the decision is actually about

Reasonably well supported. Sleeping badly makes the next day's conflict worse, measured in a 14-day diary study, a laboratory conflict task with observer coding, a 68-couple newlywed diary, a 29-couple actigraphy study and a three-study paper with a quasi-experiment — and in Gordon and Chen's data this held even when the partner was not the cause of the bad sleep. Sharing a bed costs deep sleep and adds movement: Monroe's fourteen couples in 1969 and Pankhurst and Horne's forty-six pairs in 1994 agree. People who were measured both ways preferred the worse sleep. And treating obstructive sleep apnoea improved marital satisfaction by about half a standard deviation in the one controlled comparison that measured it.

Not supported. That a third of couples sleep apart — the figure is 15% consistently and 20% occasionally, on a base of 2,005 American adults, not couples. That a third do it regularly, which is the version that ranks third. Any claim that moving rooms improves a relationship, or damages one, because no study has followed couples who did it. And any claim that it is a generational trend: the two surveys that ask disagree about which direction age runs.

Never measured at all. Couples randomised to separate rooms. Couples followed from the decision forwards. Same-sex couples, who appear in none of the studies above. And what happens over years rather than weekends.

So the evidence cannot make this decision, and pretending otherwise is the only real error available here. What it can do is reframe it. Three questions have answers you can actually get: is one of you snoring in a way that needs a doctor rather than a spare room; is the sleep loss real, which two weeks of both people noting bedtime, wake time and how the next day went will tell you better than any survey; and is this a choice the two of you are making or one that is happening to you. Chiao's loneliness finding, weak as it is, is the only measured signal in the subject, and it points at that last question rather than at the arrangement.

If what is actually in dispute is the nights rather than the sleep, that is a different problem with better evidence behind it: our conflict style test scores the demand-withdraw pattern rather than the business-meeting instruments most quizzes borrow, our relationship satisfaction test separates satisfaction from commitment, and what phones at close quarters do to two people covers the other object that ends up in the bed. For a survey misread the same way this one was — a tick-list turned into a fact about couples, with the part that complicates it left out — see in-laws and divorce. The rest of this section is under Mindset.

Pulling it together

The decision to sleep in separate rooms has no evidence base in either direction, and that is the honest headline. No study has randomised couples to separate rooms. No study has followed couples from the decision forwards. The only paper in the biomedical index that uses the phrase is a narrative review whose own closing paragraph asks somebody to go and measure the long-term effects on relationship satisfaction.

What the research does establish is narrower and more useful. Sleeping badly makes the next day's conflict worse — measured repeatedly, and in one study even when the partner was not the cause. Sharing a bed costs deep sleep and adds movement, and the people who were measured both ways preferred it anyway. Treating sleep apnoea moved marital satisfaction in the one controlled comparison that looked. And the single population-level finding about separate bedrooms is about loneliness rather than about sleep.

So the question that can actually be answered is not whether sleep divorce works. It is whether somebody in the bed has an untreated medical problem, whether the sleep loss is real enough to show up in two weeks of both of you writing it down, and whether the arrangement is one you are choosing together or one that is happening to you. Those have answers. The trend does not.

Frequently asked questions

Do a third of couples really sleep in separate beds?

No — that is a misreading of one survey. The American Academy of Sleep Medicine commissioned an online survey of 2,005 adults in the United States, fielded by Atomik Research between 24 and 29 March 2023, and asked “Do you ever do any of the following to adjust your sleep routine to accommodate a bed partner?” with six options to tick. 391 people (20%) ticked “sleep in another room on occasion” and 309 (15%) ticked “sleep in another room consistently.” The famous one-third is those two added together, on a base of all adults rather than all couples — nothing in the published table restricts the sample to people who have a bed partner. 840 respondents (42%) ticked nothing at all. The figure that describes an arrangement rather than an event is 15%.

Is a sleep divorce a sign that a relationship is in trouble?

Nobody has measured it, which is a better answer than the confident ones on either side. No study has randomised couples to sleeping arrangements, and none has followed couples from the decision forwards. The only population-level finding that exists runs mildly against the practice: Chiao and colleagues, surveying 1,645 older parents in northern Taiwan, found that married people who slept in separate bedrooms reported more emotional and social loneliness than bed-sharing couples, adjusting for sleep problems and, the authors state, independently of relationship satisfaction. That is cross-sectional, one region, and older adults, and the obvious alternative reading is that people who already feel distant are the ones who move rooms. Treat it as the only signal there is, not as a verdict.

Do you actually sleep better alone?

By objective measures, usually yes, and the research is older than the trend. Lawrence Monroe had 28 married good sleepers — 14 couples — sleep three consecutive nights in a laboratory under each of two conditions in 1969, and found significantly more Stage 4 sleep and significantly less REM when they slept alone. Pankhurst and Horne monitored 46 pairs of bed partners for eight nights with actigraphy in 1994 and found more movement during sleep when couples were together, with about a third of movements shared by both partners within the same 30-second epoch. Wendy Troxel's review summarises the whole set the same way: participants generally sleep better when sleeping alone. The twist is in the next question.

If people sleep better alone, why do they keep sharing a bed?

Because the subjective rating runs the other way, and it does so consistently. Troxel's 2010 review reports that in Monroe's study, despite the changes in sleep architecture, participants reported being less satisfied with their sleep when sleeping alone — and that Pankhurst and Horne's couples generally reported better subjective sleep quality when sleeping with their partners than when alone. Her reading is that the psychological need for closeness and security at night tends to outweigh the equally real need for good sleep. The popular framing has this exactly backwards: it assumes sleeping apart feels like a loss but is better for your sleep, when the measured position is that it is better for your sleep and feels worse.

Is it true that sharing a bed increases REM sleep?

That claim rests on twelve couples. Henning Drews and colleagues put 12 heterosexual couples aged 18 to 29 through four nights of laboratory polysomnography in 2020, two sleeping individually and two together, with the order counterbalanced, and found about 10% more REM sleep when co-sleeping, less fragmented REM, longer undisturbed REM fragments and more limb movements. No other sleep stage changed. Their own introduction notes that only two polysomnographic studies comparing co-sleep with individual sleep in healthy couples exist, and one of the two is their own pilot. Monroe's 1969 study found REM running the same direction, which is genuine corroboration. It is still twenty-four young adults who already shared a bed.

Has anyone studied couples who tried sleeping apart and then went back?

Not in the research literature. Searching PubMed for “sleep divorce” returns one record: a 2025 narrative review in the Journal of Health Psychology by Andersen, Alvarenga, Fonseca and Tufik, which discusses the practice and closes by saying future research should examine the long-term effects of different sleeping arrangements on relationship satisfaction. The only figures that exist on going back are from Sleep Foundation's own on-site survey, which reports that over a quarter of people who tried sleeping apart eventually shared a bed again and that for over a third of them missing each other was the reason. That survey publishes no sample size and no method, on a page that carries a mattress recommendation.

Does sleeping badly actually damage a relationship?

This is the part that is well supported. Amie Gordon and Serena Chen found more relationship conflict after poor nights of sleep in a 14-day daily study, and in a laboratory conflict task found one partner's poor sleep associated with a lower ratio of positive to negative affect — self-reported and observed — and with decreased empathic accuracy for both partners, with resolution most likely when both were rested. Crucially, the effects were not explained by stress, anxiety, depression, low satisfaction or by the partner being the source of the poor sleep. Maranges and McNulty found 68 newlywed couples more satisfied on days after longer sleep. What has never been shown is the missing link: that moving rooms reliably produces the better sleep.

What should we do about snoring?

Treat it as a medical question first. Loud snoring with gasping, pauses in breathing or daytime sleepiness are the cardinal symptoms of obstructive sleep apnoea, which has real cardiovascular consequences and an effective treatment. The only intervention study in this whole area is about that treatment: McFadyen and colleagues compared 44 patient-and-partner couples treated with CPAP against 25 managed conservatively and found all seven summary psychosocial scores better in the treated group at follow-up, with marital satisfaction improving by about half a standard deviation and health and sleepiness scores by more than a full one. Allocation was by position on the waiting list rather than randomised, so it is a controlled comparison rather than a trial. Moving to another room silences the noise and leaves the disorder in place.

Is sleeping apart a generational trend?

The two surveys that ask disagree about the direction. The AASM panel found sleeping in another room concentrated in the middle of the age range, with consistent sleepers-apart at 19% among 25-to-34s and 7% among 55-to-64s. The Better Sleep Council's survey reports the opposite: 16% of over-55s sleep in separate bedrooms against 3% of 18-to-34s, and it puts the overall figure at 9% rather than 35%. Neither page acknowledges the other, and the Better Sleep Council page publishes no sample size, field dates or method. The AASM sample was quota-balanced on sex to the exact person — 997 men and 997 women — but not on age, with the 35-to-44 group alone making up 27% of respondents.

Sources

What this article draws on. For the broader bibliography see our sources page.

Keep reading