Thinking About Breaking Up: What a National Panel Found Happened Next
A quarter of married people had thought about divorce in the past six months. A year later, 31% of them were no longer thinking about it.
Before the article. If you are being hurt, controlled, monitored or threatened by the person you are with, this page is the wrong frame for your situation and nothing in it should be read as a reason to stay or to wait. Find A Helpline, in the sources at the bottom, lists free and confidential services by country. In the United States the National Domestic Violence Hotline is 1-800-799-7233, or text START to 88788, 24 hours a day. The second section says why the research below does not apply there, and what does.
With that said. We put the question in the title to DuckDuckGo on 10 November 2026 and wrote down what came back, because it is what most readers will have seen before arriving here. Nine pages, and all of them answer a different question: 4 Warning Signs Your Relationship Is Already Over, Expert Lists 25 Signs, 18 Signs the Relationship Is Over for Him. Not one of the nine carries a single sample size, and none links to a journal article. Gottman's name appears on six of them as the whole evidence base. The one page that does give numbers gives them from its own product's 58,221 completed assessments — people worried enough to open a relationship test, which is not a sample of relationships.
There is a national panel that measured the thing the question is actually asking. Hawkins and six colleagues surveyed 3,000 married Americans aged 25 to 50 and then surveyed them again a year later. A quarter had thought about divorce in the previous six months, and more than half had thought at some point that their marriage was in serious trouble. A year on, 31% of the people who had been thinking about divorce were no longer thinking about it — and 36% of those who had not been, were. Only 5% of thinkers had endorsed "I'm done with this marriage"; the most common answer, at 43%, was that they did not want a divorce and were willing to work hard. About half of each type of thinker had changed type within the year, mostly toward softer thinking, and between 3.7% and 7.8% had divorced.
What follows is twelve sections: the search results and what they omit, when this is not a question about doubt, how common the thought is, what had happened a year later, how firm the thinking actually was, the three kinds of thinker, how many changed, what happened to the people who said they were done, the one study where doubt did predict, what people do while thinking, what is missing for unmarried couples, and the limits.
It does not contain an answer to "should I leave", because the research does not have one. Our editorial policy explains why every empirical claim here carries a linked source.
What the first page of results answers, and what it leaves out
The question in the title was put to DuckDuckGo on 10 November 2026, and the nine pages that came back answer a different question than the one asked.
The query asks whether the thought means anything. The results answer whether the relationship is finished: Signs Your Relationship Is Over, 4 Warning Signs Your Relationship Is Already Over, How To Know When A Relationship Is Over: Expert Lists 25 Signs, 18 Signs the Relationship Is Over for Him. Two pages from a single couples-therapy app, two from Regain, one each from Verywell Mind, Bonobology, BetterHelp, wikiHow and a therapist's practice blog. The counts are the differentiator: 4 signs, 5 signs, 18, 25.
Unlike most subjects this site has measured, this one is not abandoned by the health authorities — Verywell Mind ranks third, and it and BetterHelp are the only two pages that link out to PubMed at all. That split does not hold here, and it is worth naming because it has held in almost every other subject this site has measured.
What holds is something narrower, and it is the gap this page exists to fill. Across all nine pages there is not one sample size. No study is named with a number attached, no page carries a single link to a journal article or a DOI, and Gottman's name appears on six of the nine as the entire evidential backing. The one page that does supply figures is the app that ranks first, and its numbers come from its own product: 58,221 completed relationship assessments, of which 9,160 people selected “hopeless.” People who have opened a relationship-assessment tool are not a sample of relationships — they are a sample of people worried enough to open one. It is the same problem as the success rates in what has actually been measured about cutting contact with an ex, where every published figure was calculated on a company's own paying clients.
And the framing itself is the omission. Every page treats the thought as a symptom to be decoded against a checklist, which assumes the thought holds still long enough to decode. There is a national panel that measured whether it does, and no page on the first results page mentions it.
Before the research: when this is not a question about doubt
If you are being hurt, controlled, monitored or threatened by the person you are with, nothing on this page applies to your situation, and none of it should be read as a reason to stay or to wait. Find A Helpline, in the sources below, lists free and confidential services by country. In the United States the National Domestic Violence Hotline is 1-800-799-7233, or text START to 88788, 24 hours a day. They will plan with you whether or not you have decided anything.
This is at the top rather than the bottom because the research below is about ambivalence, and ambivalence is the wrong frame for danger. The whole finding of this article is that uncertainty about a relationship is common, moves around, and does not forecast an ending. Applied to someone whose partner frightens them, that is not reassurance, it is a delay — and the period around leaving is the one that needs planning with a service rather than an article.
Control is describable rather than intuitive, and naming it is more useful than naming a feeling: deciding who you may see, monitoring your time and requiring an account of it, controlling money or transport, going through your phone, pressure that continues after a no, punishment by silence or by anger. Where those are present, what is sometimes offered online as an explanation for staying is examined in what trauma bonding actually rests on, and the psychology of jealousy works through telling the feeling apart from the behaviour.
The same applies, differently, to distress in the person reading. If the thinking has stopped being thinking and turned into not sleeping, not eating or thoughts of harming yourself, that is a reason to talk to a doctor rather than to keep weighing a relationship. The helplines above answer that too.
With that said, the rest of this page is about the ordinary case: someone who keeps turning the question over and wants to know what it means.
The question has been put to 3,000 married people, twice
The useful study here is not about signs. It is about the thought, and the people who have it were asked about it twice.
Alan Hawkins, Adam Galovan, Steven Harris, Sage Allen, Sarah Allen, Kelly Roberts and David Schramm surveyed 3,000 married people aged 25 to 50 in early 2015 and came back 12 to 13 months later. Everyone had been married at least a year. The age band was chosen deliberately to load the sample with parents of children still at home: 80% were parents, the average number of children was 1.93, and 66% still had children in the household. Average marriage length was 12.04 years, average age 39.3, and 53% were women.
The prevalence figures are the first thing worth having, because they reframe the question before any analysis starts.
Fifty-three per cent of them said yes to have you ever thought your marriage was in serious trouble? More than half. Of those, 28% had had such thoughts in the past but not within the last six months, and another 25% had thought about divorce within the last six months.
So at any given moment a quarter of this sample was thinking about it, and over half had been there at some point. That alone answers part of the title: whatever the thought is, it is not rare, and it is not a minority condition belonging to marriages that are visibly failing.
There is one more figure from that past-but-not-current group, and it is the most quietly encouraging number in this literature. Asked how they felt about it now, 88% of them said they were glad they were still married. Less than one per cent — 0.5% — said they were not glad. The remaining 11.5% reported mixed or unsure feelings. These are people who had thought their marriage was in serious trouble, were no longer thinking it, and looked back without regret.
Two honest notes on what kind of study this is. The data were collected through YouGov's online panel, then matched and weighted against a sampling frame built from the 2010 American Community Survey, which the authors say produced data that “closely approximated a nationally representative sample.” That is a matched and weighted opt-in panel, not a probability sample, and it is a real difference even if a careful one. And because everyone in it was married at the time, the sample by construction excludes people whose marriages had already ended — the authors flag this truncation themselves.
What had happened a year later
This is the finding, and it is the one the first page of search results has no equivalent of, because it requires asking the same people twice.
Across the whole sample, a year on: 93% were still married, 4% had separated and 2% had divorced.
Now split it by what people had been thinking at the start.
Among those who were thinking about divorce at Time 1, 31% reported no recent thoughts about it a year later. Sixty-nine per cent were still thinking about it — a figure that includes the 4% who had separated and the 3% who had divorced or remarried.
And the mirror result, which is the one that stops this being a story about relief: among those who were not thinking about divorce at Time 1, 36% were thinking about it a year later. That group includes 2.5% who were divorced or remarried by the second survey, which the authors note as evidence that dissolution does sometimes arrive quickly.
Read those two numbers together and the thought stops looking like a verdict and starts looking like weather. Roughly a third of the people who had it put it down within a year. Roughly a third of the people who did not have it picked it up. The authors' own summary of their paper is that divorce ideation “is common but dynamic, and it is not necessarily an indication of imminent marital dissolution.”
The asymmetry matters for how the number should be used. A reader arriving at this page is, by definition, in the first group — and for that group the measured outcome a year later was overwhelmingly still married and, in about a third of cases, no longer thinking about it. That is not a prediction about any particular relationship, and the next sections are about why it cannot be one. It is a correction to a base rate that the genre states as though it ran the other way.
Worth separating from ordinary decline: satisfaction drifts slowly downward in most couples and that is a different phenomenon from the one measured here, as the trajectory research behind the honeymoon phase sets out.
Most “thinking about divorce” is not what the phrase implies
The phrase does a lot of work that the data do not support. When the 745 recent thinkers were asked how often and how firmly, most of them turned out to be somewhere much earlier than the words suggest.
On frequency: 17% of the full sample had thought about divorce only a few times in six months, 5% several times, and 3% a lot of times. Within the thinkers, that means 70% had infrequent thoughts — a few times across half a year.
On firmness, the authors offered a set of statements and asked people to pick one. The distribution is the thing:
- Only 5% chose “I'm done with this marriage.” Most of the few who did had been thinking about it frequently.
- The modal answer, at 43%, was “I don't really want a divorce; I'm willing to work hard to keep us together.”
- 23% reported mixed feelings about a divorce.
- Another 23% said they would consider working on the marriage rather than divorcing if their spouse got serious about making some major changes.
- 6% said none of the statements matched.
So the overwhelming majority of people who would truthfully answer yes to are you thinking about divorce were, in the same breath, describing openness to repair. One in twenty had concluded anything.
That is worth holding against the checklist framing of the search results. A page that reads the presence of the thought as a sign the relationship is over is treating the 5% case as the default and the 43% case as denial.
Two demographic details, both small. Women were somewhat more likely than men to report recent thoughts, 27% against 22%. And the thought becomes less common the longer a marriage lasts — steady at roughly a quarter through the first fifteen years (26% at 1–5 years, 24% at 6–10, 23% at 11–15), then falling to 15% at 16–20 years and 12% beyond 21. Other demographic differences between thinkers and non-thinkers were, in the authors' words, few and small.
What distinguishes relationships that last is a separate question with its own measurements, and the markers are more specific than a feeling: what the research on couples who stay together reports.
Three kinds of thinking about it, and the odd one out
Averages hide shapes. Rather than treat the 745 thinkers as one group, the authors ran a latent class analysis on all of the ideation measures at once and asked whether distinct kinds of thinker fell out of the data. Three did.
Soft thinkers — 49%. The largest group. Relationship hope averaged 5.94 on a seven-point scale, connection problems 2.42 out of six. Ninety-one per cent had thought about divorce only a few times in six months, and 28% had been thinking about it for three months or less.
Long-term-serious thinkers — 45%. Hope 4.29, connection problems 4.12. Forty-six per cent had been thinking about divorce for two years or more, and 21% thought about it a lot of times. This is the group the phrase in the title is usually imagined to describe, and it is slightly under half of the thinkers.
Conflicted thinkers — 6%. The small group, and the interesting one, because it breaks the pattern the other two establish. Their problems were the most severe of the three — connection problems 5.51 of six, intense problems 3.28 of four, and 81% named mental health as a major reason — and yet their relationship hope was the highest of all three groups at 6.01, marginally above the soft thinkers. They were also about six years younger than the other thinkers and more religiously devout.
That combination is the most useful thing in the classification, because it is the one a checklist cannot represent. Severity of problems and hope for the relationship are not two ends of one dial. A person can be in the worst-rated marriages in the sample and hold the most hope in the sample at the same time, and the authors speculate that the devoutness is part of how both fit together — more intense problems, more likelihood of thinking the marriage is done, and still conflicted about divorce and hopeful about the future.
There is a parallel worth noting rather than restating: in a clinical sample of 913 couples, 40.3% fell into a class defined by the two partners being out of step on commitment and on hope, which is the subject of what the research says when one partner won't go to couples therapy. Hope turns out to be a measured variable in both places, and in both it refuses to track severity.
About half of them had changed group within a year
Having established three kinds of thinker, the authors did the thing that makes the paper worth reading: they checked who was still in the same group a year later. A latent transition analysis produced the probability of each Time 1 class landing in each Time 2 class, with non-thinking and divorced added as destinations.
The instability is substantial, and it runs in one direction more than the other.
Soft thinkers (49% of thinkers) a year later: 43.9% still soft thinkers, 41.2% had become non-thinkers, 10.2% had become serious thinkers, 0.9% conflicted, and 3.7% had divorced.
Long-term-serious thinkers (45%): the most stable group, with 50.7% still there — but 20.7% had softened to soft thinkers, 21.4% had become non-thinkers, 2.2% conflicted, and 4.9% had divorced.
Conflicted thinkers (6%): 49.0% still conflicted, 25.7% had become non-thinkers, 17.5% had become serious thinkers, 0.1% soft, and 7.8% had divorced — the highest divorce rate of the three, which is what you would expect from the group with the most severe problems.
Taken together, roughly half of each group had moved: 56% of soft thinkers, 49% of long-term-serious thinkers and 51% of conflicted thinkers were somewhere else a year on, and the movement was mostly, in the authors' phrase, “in the direction of less and softer thinking about divorce.”
The divorce numbers deserve to be read carefully, because they are the ones a reader will want. Between 3.7% and 7.8% of people who were actively thinking about divorce had divorced one year later, depending on which kind of thinker they were. That is a one-year window, not a lifetime risk, and over longer horizons the figure would be larger — this study cannot say how much larger. What it does say is that the thought and the outcome are separated by a great deal of time and a great deal of change.
One methodological point in the study's favour, since attrition is the obvious way a result like this could be manufactured. The authors tested whether some classes of thinker were more likely to drop out before the second survey and found no significant difference (χ²(2) = 0.73, not significant), so the people who left the study were retained in the transition analysis through imputation rather than quietly dropped. Overall retention was 80% of reachable participants and 75% of the full sample; there was no difference in divorce-thought frequency between those retained and those lost, and a slight one in marital happiness, with less happily married people marginally more likely to be retained.
Even “I'm done with this marriage” moved
The transition analysis is about classes. The authors also looked at the plainer question of whether people's stated attitude held still, and it is the sharpest version of the finding.
Only one-third of Time 1 thinkers reported the same attitude about divorce a year later. Two-thirds had changed their answer within twelve months.
And then the subgroup that matters most for anyone reading this page while fairly certain. Among the small number who had selected “I'm done with this marriage” at Time 1, just 29% reported the same attitude one year later. Fifty-three per cent reported attitudes suggesting more openness to staying married — including 18% who reported no recent thoughts about divorce at all.
That is a strong result and it should be stated with its weaknesses visible, because it is also the most fragile number in the article. The group is small: 5% of 745 thinkers, so on the order of thirty-odd people, and the paper itself calls it “the small number of thinkers.” Percentages of a group that size move a long way if a handful of people answer differently. It is reported here because the direction is consistent with everything else in the study, not because the figure is precise.
What it supports is narrow and worth saying anyway. A settled-feeling conclusion about a relationship is not a measurement of the relationship; it is a measurement of a person on the day they were asked. The same asymmetry — a feeling that presents itself as information about the future and turns out to be information about the present — is the best-documented result in the adjacent literature on forecasting a breakup, set out in why people stay in relationships that are not making them happy.
None of which means a person's sense that something is wrong should be overridden. Two-thirds of attitudes changed; one-third did not, and the long-term-serious thinkers were the most stable group in the data. The claim is about the reliability of the thought as a forecast, not about whether it is worth taking seriously.
Where doubt does predict: the research on cold feet
An article that stopped at the previous section would be doing the same thing as the pages it criticises, in the opposite direction. There is a well-designed study in which doubt about a relationship predicted exactly what people fear it predicts, and leaving it out would be dishonest.
Justin Lavner, Benjamin Karney and Thomas Bradbury asked 464 recently married spouses whether they had ever been uncertain about getting married, and then followed them, comparing four-year divorce rates and satisfaction trajectories between those with and without premarital doubts.
Doubts were common: at least one partner reported them in two-thirds of couples.
They were also not harmless. Women with premarital doubts had significantly higher four-year divorce rates — and that held with concurrent marital satisfaction controlled, along with the difficulty of the engagement, parental divorce history, premarital cohabitation and neuroticism. Among the couples who were still together, both men's and women's doubts predicted less satisfied marital trajectories. The authors' conclusion is the useful phrase: premarital doubts are “common but not benign.”
So the two studies are not in conflict, and the distinction between them is the practical content of this page. They measure doubt at different moments about different things. Lavner and colleagues measured uncertainty before a marriage began, about whether to enter it, in people who then entered it anyway. Hawkins and colleagues measured thoughts about ending a marriage already a decade old on average, in a population where the thought is common and recurrent.
The honest synthesis is that the predictive value of doubt depends on which question the doubt is answering. Uncertainty about whether to commit at all, recorded at the point of commitment, carried real signal four years out. Recurrent thoughts about ending an established marriage, measured against what happened one year later, mostly did not — they moved, and they mostly moved toward softening.
A second honest note: both of these are findings about averages in samples of mixed-sex American couples, and the gendered part of the Lavner result is a single finding in a single sample. It is reported as the authors reported it, not as a law about who should trust their doubts.
What people actually do while they are thinking about it
If the thought is common and unstable, the next question is what happens in the year between surveys, and the same project measured that too.
Adam Galovan, Hawkins, Harris and David Simpson analysed the repair behaviour of the 745 people who were thinking about divorce, and sorted them by what they actually did. Four classes came out:
- Intense Seekers — 6%. High levels of every kind of repair behaviour, professional services included.
- Moderate-fading Seekers — 14%. Moderate levels of various repair behaviours including professional services, but they did not sustain it across the year.
- Minimal-private Seekers — 42% and Private-sustained Seekers — 38%. Both avoided professional services entirely, engaging in low-to-moderate personal and private efforts.
The headline is in that last pair. Eighty per cent of people thinking about ending their marriage used no professional help at all. The authors' own conclusion is that the field needs to pay more attention to personal and self-help approaches, because that is how most people actually try to repair a relationship.
The first study fills in how private this is. Of the recent thinkers, about 40% had talked to their spouse about their thoughts, about 40% had not, and 20% chose “Maybe, we sort of talked about it” — the paper's own wording, which the authors read as vague or indirect conversations that may never have mentioned divorce. And only 9% had spoken to a lawyer or mediator.
So for a large share of these marriages, the thinking is happening in one head. The authors call divorce ideation “often an intrapersonal process,” and name it as a limitation of their own design: they surveyed one spouse, so nobody knows what the other one was thinking.
There is a small qualitative companion to this worth one line, because it points at something the numbers cannot. Shauna Fenske, Harris and colleagues interviewed 30 people in the middle of a divorce decision and re-interviewed 22 a year later, then analysed what they said about the effect of the first interview. Three themes came out: talking got me thinking, thinking got me acting, and the conversation was — surprisingly to the participants — therapeutic. Thirty people is thirty people and this is thematic analysis, not measurement. But it is at least consistent with the rest: the thought is not a fixed object being reported on, and the act of articulating it to someone appears to change it.
On the professional side of that, this site has two companion pages with the outcome data: what the trials on couples therapy actually report, which includes the evidence that greater distress at intake predicted larger gains rather than smaller, and what the research says when one partner won't go. Where the problem is a specific recurring fight rather than the relationship as a whole, the evidence on repair after conflict is the narrower piece, and the conflict style test scores how a couple argues as a separate thing from how satisfied they are.
If you are not married, the evidence thins out fast
This needs saying plainly, because the query in the title is usually typed by someone who is not married, and every figure above comes from married people.
The Hawkins panel is married Americans aged 25 to 50, married at least a year, averaging twelve years in, mostly with children at home. Nothing in it was measured on dating or cohabiting couples, and the mechanics are not obviously transferable: the thought is cheaper to act on without a marriage, a mortgage and children, and the paper's own finding that ideation declines with marriage length suggests duration is doing something. The step before marriage has its own literature and its own arguments — what the evidence on living together before marriage actually reports covers the four conditions that change the answer, including the one that moves a hazard ratio from significant to absent.
What exists for unmarried couples is thinner and this site has already set it out rather than restate it here. The one prospective study that recruited people actively contemplating a breakup followed 500 of them for two months and found that 29% had ended the relationship by the follow-up — and, more interestingly, that how satisfied they were did not predict which ones. That study, with its sample and its numbers, is the spine of why people stay in relationships that are not making them happy.
The closest thing to a stability figure outside marriage is also covered elsewhere: among 792 young adults, 44.8% of those who broke up got back together without any technique involved, which is the base rate that makes claims about reunion methods unfalsifiable — what has actually been measured about cutting contact with an ex sets that out, along with the finding that the couples who broke up and reunited reported physical violence at twice the rate of those who did not.
Put together, the honest position is this. The claim that thinking about ending it does not reliably predict ending it is well measured in marriages and plausible but not established in dating relationships. Anyone applying the figures above to a two-year relationship is extrapolating, and should know they are.
One timescale worth keeping separate: if part of what is being weighed is whether your own patterns would follow you out, the shifts the evidence supports there are measured in years rather than months and are easy to confuse with a bad fortnight — what the longitudinal evidence says about attachment styles changing. And if the relationship has already ended and the question is whether you are over it, that is a different literature with a different marker: the signs you have moved on and the signs you have not.
What this evidence cannot tell you
The limits go here rather than in a footnote, because on a page someone may be reading while deciding something, they are part of the answer.
It cannot tell you whether your relationship is over. Not approximately, not probabilistically. Every figure above describes what proportion of a group was somewhere a year later, and a proportion is not a probability about one person. The study was designed to map a phenomenon, and its authors are explicit that their contribution is “the empirical contours of divorce ideation” — not a forecast and not a decision rule.
Nobody asked the other person. The authors name this first among their own limitations: they surveyed one spouse. Given that 40% of thinkers had not told their partner and another 20% had only “sort of” talked about it, the dyadic half of this — what both people are thinking, and what happens when the thought is finally said out loud — is simply unmeasured. The 2026 work on couples who arrive out of step is the nearest thing, and it is a clinical sample rather than a national one.
It is one year, in one country, in one decade. The panel ran from early 2015 to 2016 among US adults. A year is short for a question people often sit with for much longer, and the 3.7%–7.8% divorce figures would be larger over five years by an amount this design cannot estimate. The authors also restricted the sample to married 25-to-50-year-olds with the explicit aim of capturing parents, so it is not a general portrait of marriage.
It is a matched and weighted opt-in panel, and it is self-report. The weighting against an American Community Survey frame was done carefully and the authors claim only that it “closely approximated” national representativeness. Beyond that, everything here is what people said about their own thoughts in a nine-minute survey, twice. The authors' second stated limitation is that they measured cognition and behaviour and did not explore the emotional landscape at all.
And the thing that moved may be the question, not the marriage. A person who reports no recent thoughts about divorce at Time 2 may have resolved something, or settled for something, or simply had a quieter six months before the survey arrived. The panel measures the thought, not what the thought was tracking — and the one study that followed people whose problems were best documented, the conflicted thinkers, found both the highest hope and the highest divorce rate in the same small group.
What survives all of that is modest and still worth the reading time. Thinking about ending a relationship is common, most of it is infrequent and open to repair rather than settled, it moves a great deal within a year and mostly moves toward softening, and among the married people who had been there and come out the other side, 88% were glad they were still married. None of that is a reason to stay and none of it is a reason to go. It is a reason not to treat the thought itself as the answer.
If the question has already moved past the thinking and into the aftermath, the figures circulating for that stretch turn out to have the same problem this page started with — a confident number with no document behind it — and what has actually been measured about how long it takes to get over a breakup follows each of them back to its source.
Every empirical claim on this page carries a linked source, which is the promise made in our editorial policy; claims this site has had to retract are listed on the corrections page. Which of satisfaction, alternatives and investment is holding a particular relationship up is a question with an answer, and it is the one our relationship satisfaction test is built to print rather than to judge. The rest of this section is under Mindset.
Pulling it together
The short version is that the thought is common, unstable, and a poor forecast of the outcome — and that none of this is advice about what to do.
More than half of 3,000 married Americans had at some point thought their marriage was in serious trouble, and a quarter had thought about divorce in the previous six months. Within that quarter, 70% of the thinking was infrequent, only 5% had concluded they were done, and 43% said outright that they did not want a divorce and were willing to work. A year later, 31% were no longer thinking about it, about half of each type of thinker had shifted type, the movement was mostly toward softening, and between 3.7% and 7.8% had divorced. Among the people who had been there in the past and were no longer thinking about it, 88% said they were glad they were still married.
Two things keep that from being reassurance. Doubt measured at the point of committing to a marriage did predict divorce four years out, in a sample of 464 spouses — common but not benign. And 36% of the people who were not thinking about divorce at the first survey were thinking about it at the second, so the instability runs both ways.
What the research supports is narrow: a recurring thought about ending an established relationship is not in itself evidence that the relationship is ending. It is also not nothing. The one thing nobody measured is what the other person is thinking, and in this data 40% of thinkers had never said a word about it.
Frequently asked questions
Does thinking about breaking up mean it's over?
Not on the best available measurement. Hawkins and colleagues surveyed 3,000 married Americans aged 25 to 50 and resurveyed them a year later. Among those thinking about divorce at the first survey, 31% reported no recent thoughts about it one year on, and between 3.7% and 7.8% had divorced, depending on which of three types of thinker they were. Across the whole sample, 93% were still married a year later. The authors' own conclusion is that divorce ideation "is common but dynamic, and it is not necessarily an indication of imminent marital dissolution." Two caveats worth holding. The instability runs both ways: 36% of people who were not thinking about divorce at the first survey were thinking about it at the second. And everyone in this study was married for at least a year, averaging twelve, so none of it was measured on dating couples.
How common is it to think about ending your relationship?
Commoner than the genre implies. In the Hawkins panel of 3,000 married people aged 25 to 50, 53% said yes to "have you ever thought your marriage was in serious trouble?" Twenty-eight per cent had had such thoughts in the past but not in the last six months, and another 25% had thought about divorce within the last six months. Within that quarter, the thinking was mostly infrequent: 17% of the full sample had thought about it only a few times in six months, 5% several times and 3% a lot of times — so about 70% of thinkers had infrequent thoughts. Women reported it slightly more often than men, 27% against 22%, and it became less common with marriage length, holding near a quarter through the first fifteen years before falling to 15% at 16–20 years and 12% past 21.
Do people who think about divorce regret not leaving?
In the one dataset that asked them, overwhelmingly not. Of the 28% of Hawkins's sample who had thought their marriage was in serious trouble in the past but not within the last six months, 88% said they were glad they were still married. Half a per cent — 0.5% — said they were not glad, and 11.5% reported mixed or unsure feelings. Two things to hold against that figure. It is retrospective self-report, and people have good reasons to narrate a resolved crisis as having worked out. And it is measured only among people who were still married, so by construction it cannot tell you anything about how the people who left feel about leaving; the research that compared the two outcomes found roughly half of divorcing couples came out happier and half worse off, which is covered in why people stay in relationships that are not making them happy.
Are there different kinds of thinking about divorce?
Three, in the data. A latent class analysis of the 745 recent thinkers produced soft thinkers (49%), with relationship hope averaging 5.94 of 7 and 91% thinking about it only a few times in six months; long-term-serious thinkers (45%), with hope at 4.29 and 46% having been thinking about it for two years or more; and conflicted thinkers (6%), who had the most severe problems in the sample — connection problems 5.51 of 6, and 81% naming mental health as a major reason — and yet the highest relationship hope of all three groups at 6.01. That last group is the useful one, because it shows severity and hope are not two ends of one dial. Conflicted thinkers were also about six years younger and more religiously devout than the other two groups.
If I'm certain I want to leave, does that change?
Sometimes, and more often than people expect, though this is the weakest number on the page. Only one-third of Hawkins's Time 1 thinkers reported the same attitude about divorce a year later. Among the small group who had selected "I'm done with this marriage" at the first survey, just 29% reported the same attitude a year on; 53% reported attitudes suggesting more openness to staying married, including 18% with no recent thoughts about divorce at all. The weakness is the group size: that was 5% of 745 thinkers, so a few dozen people, and percentages of a group that size move easily. It is reported because the direction matches everything else in the study. It is not a reason to override your own judgement — the long-term-serious thinkers were the most stable group in the data, with 51% unchanged.
Are doubts about a relationship ever a real warning sign?
Yes, and the clearest case is doubt measured at the point of commitment rather than years into it. Lavner, Karney and Bradbury asked 464 recently married spouses whether they had ever been uncertain about getting married. Doubts were reported by at least one partner in two-thirds of couples — so they are normal — but women with premarital doubts had significantly higher four-year divorce rates, holding even after controlling for current marital satisfaction, the difficulty of the engagement, parental divorce, premarital cohabitation and neuroticism. Among couples still together, both partners' doubts predicted less satisfied trajectories. The authors call premarital doubts "common but not benign." The distinction from the divorce-ideation findings is what the doubt is about and when: uncertainty about entering a marriage carried four-year signal, while recurring thoughts about ending an established one mostly moved and mostly softened.
Should I tell my partner I've been thinking about breaking up?
The research does not answer that, and anyone who says it does is going beyond the evidence — but it does describe how unusual it is to have said anything. Among recent thinkers in the Hawkins panel, about 40% had talked to their spouse about their thoughts, about 40% had not, and 20% chose "Maybe, we sort of talked about it," which the authors read as indirect conversations that may never have mentioned divorce. Only 9% had talked to a lawyer or mediator. The authors describe divorce ideation as "often an intrapersonal process" and name as their own first limitation that they surveyed only one spouse, so what happens after the thought is spoken is genuinely unmeasured. The one adjacent hint is qualitative and small: 30 people interviewed mid-decision reported that talking about it changed their thinking, and found the conversation unexpectedly therapeutic. That is thirty people and a thematic analysis, not a finding.
Where do most people go for help when they're thinking about divorce?
Nowhere professional, in the great majority of cases. Galovan, Hawkins, Harris and Simpson sorted the 745 thinkers by what they actually did over the year and found four groups: Intense Seekers (6%), who used every kind of repair behaviour including professional services; Moderate-fading Seekers (14%), who used some but did not sustain it; and Minimal-private Seekers (42%) and Private-sustained Seekers (38%), who avoided professional services altogether. That is 80% using no professional help at all, and only 9% of thinkers in the first study had spoken to a lawyer or mediator. The authors' conclusion is that the field should pay more attention to personal and self-help approaches, because that is how most people actually try. On whether the professional route works when people do take it, what the trials on couples therapy report covers the outcome data, including the finding that greater distress at intake predicted larger gains rather than smaller.
Does any of this apply if we're not married?
Not directly, and it is worth being clear about that. Every figure here comes from people married at least a year, averaging twelve years, aged 25 to 50 and mostly with children at home — and the same study found divorce ideation declining with marriage length, which suggests duration matters. For unmarried couples the evidence is thinner. The one prospective study of people actively contemplating a breakup followed 500 of them over two months and found 29% had ended the relationship, with satisfaction not predicting which ones; it is set out in why people stay in relationships that are not making them happy. And among 792 young adults, 44.8% of those who broke up got back together, covered in what has actually been measured about cutting contact with an ex. So the claim that the thought does not reliably predict the outcome is well measured in marriages and merely plausible outside them.
Sources
What this article draws on. For the broader bibliography see our sources page.
- Hawkins, A. J., Galovan, A. M., Harris, S. M., Allen, S. E., Allen, S. M., Roberts, K. M. & Schramm, D. G. (2017) — What Are They Thinking? A National Study of Stability and Change in Divorce Ideation. Family Process, 56(4), 852-868 — 3,000 married Americans aged 25-50, surveyed twice a year apart; the source of every prevalence, class and transition figure on this page
- Hawkins, A. J. and colleagues (2017) — Open-access author copy of the same paper, with the tables quoted here (Tables 2.1, 2.2, 3 and 4) — deposited at Utah State University
- Galovan, A. M., Hawkins, A. J., Harris, S. M. & Simpson, D. M. (2022) — What are they doing? A national survey of help-seeking and relationship-repair behavior of individuals who are thinking about divorce. Journal of Marital and Family Therapy, 48(2), 371-390 — the same 745 thinkers; 80% used no professional services
- Lavner, J. A., Karney, B. R. & Bradbury, T. N. (2012) — Do cold feet warn of trouble ahead? Premarital uncertainty and four-year marital outcomes. Journal of Family Psychology, 26(6), 1012-1017 — 464 recently married spouses; doubts were common but not benign
- Doherty, W. J., Harris, S. M. & Wilde, J. L. (2016) — Discernment Counseling for "Mixed-Agenda" Couples. Journal of Marital and Family Therapy, 42(2), 246-255 — 100 consecutive cases, about half of whom chose couples therapy in order to reconcile
- Fenske, S. R., Harris, S. M., LaPlant-Braughton, J., Allen, S., Roberts, K. & Anderson, G. (2024) — It got me thinking: The impact of participating in a divorce decision-making interview. Journal of Marital and Family Therapy, 50(4), 867-881 — 30 people interviewed mid-decision, 22 re-interviewed a year later; thematic analysis
- Joel, S., Impett, E. A., Spielmann, S. S. & MacDonald, G. (2018) — How interdependent are stay/leave decisions? On staying in the relationship for the sake of the romantic partner. Journal of Personality and Social Psychology, 115(5), 805-824 — the prospective study of 500 people actively contemplating a breakup, referenced here for the unmarried comparison
- Find A Helpline — Free, confidential support services by country
- National Domestic Violence Hotline (United States) — 24/7 free and confidential support: 1-800-799-7233, or text START to 88788
Keep reading
- MindsetDoes a Sleep Divorce Work? The One-Third Figure Counts Adults, Not CouplesA 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.
- MindsetIn-Laws and Divorce: Half of the Famous 20% Is in the Paper, Half Was Never ThereA husband close to her parents: 20% lower divorce odds, published. A wife close to his: that figure is in no paper, and the real result is about race.
- MindsetBest Age to Get Married: The Famous 28-to-32 Window Came From a Blog PostThe 11%-a-year figure comes from a 2015 blog post, and its own model turns at 28, not 32. In the same survey, in a journal, the curve flattens instead.
- DatingDo Friends-to-Lovers Relationships Last Longer? The Two-Thirds Figure Counts How Couples Met, Not How They EndedTwo thirds of couples started as friends. That counts how relationships begin, not how long they last, and the only direct comparison is unpublished.
- MindsetMoney Arguments and Divorce: The "Top Predictor" Line Came From a Press ReleaseThe famous line is a quote in a 2013 press release. The paper is paywalled, and a co-author's other 2012 paper calls money only mildly predictive of divorce.
- MindsetSplitting Housework and Divorce: The Famous 50% Comes From a Report Almost Nobody Has ReadThe report behind "couples who share chores divorce 50% more" is 228 pages in Norwegian. Read in full, the gap is 3.1% against 4.6% over four years.