Will my relationship last? The signals research has found (and what they don’t tell you)

Can you really predict whether a relationship will last?

You may have read that John Gottman could predict a divorce “with 90% accuracy” by watching a couple talk for fifteen minutes. That’s half true, and the missing half matters. In their studies from the 1990s, Gottman and Levenson did sort couples filmed in the lab into those who would split up and those who would stay together, with impressive accuracy. But that accuracy was calculated after the fact, on the same couples used to build the formula. When you apply a formula to couples it has never seen, it gets things wrong much more often. Several researchers, Richard Heyman among them, have pointed this out: these are classifications, not prophecies.

The takeaway isn’t “we know nothing.” It’s the opposite: the factors these studies uncovered are solid, they have been found again in dozens of studies since, and they have one precious property. They are behaviors, not personality traits. So they can change. A warning sign isn’t a sentence, it’s an address: the place to start working.

The signals that weigh the most, in both directions

Contempt, and the other three horsemen

Of everything Gottman measured, contempt is the strongest predictor of separation: eye rolls, sneering, sarcasm, talking about your partner as if they were beneath you. It comes with three companions that make up the four horsemen: criticism that attacks the person, defensiveness that shifts the blame, and stonewalling. Their presence matters, but what matters even more is how often they show up and what happens afterward. A couple who lets one horseman slip into an argument and apologizes ten minutes later is doing fine. A couple where contempt is the normal tone at the dinner table is not.

Repair, and above all how it’s received

Every couple says things they regret. The difference is what Gottman calls repair attempts: a joke in the middle of an argument, a “wait, let me start over,” an outstretched hand. And even more than that: whether the other person accepts them. In his observations, the couples who last don’t argue less, they have repairs that land. If, in your relationship, every attempt gets shut down (“oh, sure, play the clown”), that’s a more worrying sign than the argument itself.

The 5-to-1 ratio during conflict

Gottman and Levenson measured, second by second, the positive and negative interactions of couples during a disagreement. Among stable couples, there were about five positive moments for every negative one, even in the middle of a conflict: a smile, a nod, a concession, a touch of humor. Among couples who went on to split up, the count fell to about one to one. You can’t measure this ratio in a single evening, but you can ask yourself honestly: in your arguments, are there still gestures of connection?

The way you tell your story

This is the most surprising signal. In 1992, Buehlman, Gottman and Katz simply asked couples to tell their story: how they met, the early days, the hard times. Three years later, the couples who had told it with affection for each other, with “we” rather than “I,” and who described the tough moments as something they got through together, were almost all still together. The couples who had rewritten the past in a negative light (“actually, from the very start…”) had, for the most part, split up. The past doesn’t change, but the way a couple tells it shows where they stand today.

Perceived commitment, and gratitude

In 2020, the study by Samantha Joel and 85 co-authors, covering more than 11,000 couples, ranked the predictors of relationship satisfaction. At the top: the conviction that your partner is committed to the relationship. Right behind it came expressed appreciation, satisfaction with intimacy, the belief that your partner is happy with you, and the way you handle conflict. And one conclusion that challenges conventional wisdom: each person’s individual characteristics taken separately predict much less than what the couple builds together. So the question “will we last?” is less about “are we right for each other?” than about “what are we doing for each other?”

Turning toward each other in the small moments

In his study of newlyweds, Gottman counted what he calls “bids for connection,” the small everyday requests for attention: “look at that bird,” a sigh, an offhand question. Six years later, the couples who were still together had responded to those bids nearly nine times out of ten; the couples who had divorced, about one time in three. There’s nothing spectacular about it. That’s exactly the point: a relationship isn’t made in grand declarations but in hundreds of tiny responses every week.

What doesn’t predict much, contrary to popular belief

• How often you argue. Stable couples don’t argue less. Gottman even described “volatile” couples who argue a lot, loudly, and last, because they repair just as loudly.
• Personality differences. “We’re too different” is one of the most common things people say, and one of the weakest predictors in Joel’s study. Two different people who treat each other well do better than two twins who hold each other in contempt.
• The early passion fading. Elaine Hatfield distinguishes passionate love, intense and unstable, from companionate love, calmer and deeper. Moving from one to the other is the normal path, not a sign of the end. What matters is what replaces the passion, not that it fades.
• Not sharing the same interests. That makes weekends easier, but it doesn’t hold a relationship together. What does is being interested in what interests your partner, which is not the same thing.

“Will my relationship last quiz”: what these quizzes are worth

The top results for this search are ten-question quizzes, or calculators that spit out a percentage from two first names. The calculators are a game, and they know it. The quizzes are often a list of more or less well-chosen signs with a verdict at the end. The problem isn’t that they’re wrong, it’s that they answer the wrong question. A serious test doesn’t say “yes” or “no.” It tells you where things hold up and where they leak: communication is solid but there’s no repair after arguments; trust is there but intimacy has gone quiet. That map is what’s useful, because you can act on it. Our own test works this way, and we say so plainly: it doesn’t predict whether a couple will last, and its score is not a scientific validation of your relationship. It’s a snapshot, taken with the right instruments.

What do you do with the answer?

If you recognized some good signs, great, and that’s no reason to do nothing: the couples who last are the ones who maintain it, not the ones who got a good start. If you recognized some bad signs, read the list again: each one is a behavior, and each one has a known antidote. Criticism gets replaced by a specific complaint. Defensiveness by taking a share of the responsibility. Contempt by a culture of appreciation, said out loud. Stonewalling by an announced pause and a set time to come back to it. We go through them in detail in the article on arguments. And the story signal, the one about your history, you can test tonight.

And if the signals are bad?

No signal, not even contempt, is a verdict: couples come back from a long way off when both people commit to it. What should raise a flag is the accumulation, and the duration: entrenched contempt, no repair that lands for months, and one of you who has already started leaving in their head. At that point, the question is no longer “will we last,” it’s “will we seek help now or in a year.” A couples therapist or a relationship counselor works on exactly these signals. And if what you feel isn’t worry but fear, if you walk on eggshells, if you’re being monitored or put down, the question is no longer the one this article is about: read the page on toxic relationships, it was written for that, and in France, 3919 answers for free, day and night. Elsewhere, contact your local domestic violence helpline.

Frequently asked questions

References

1. Gottman, J. M., & Levenson, R. W. (1992). Marital processes predictive of later dissolution: Behavior, physiology, and health. Journal of Personality and Social Psychology, 63(2), 221-233.
2. Gottman, J. M., & Levenson, R. W. (2000). The timing of divorce: Predicting when a couple will divorce over a 14-year period. Journal of Marriage and Family, 62(3), 737-745.
3. Heyman, R. E., & Slep, A. M. S. (2001). The hazards of predicting divorce without crossvalidation. Journal of Marriage and Family, 63(2), 473-479.
4. Buehlman, K. T., Gottman, J. M., & Katz, L. F. (1992). How a couple views their past predicts their future: Predicting divorce from an oral history interview. Journal of Family Psychology, 5(3-4), 295-318.
5. Joel, S., et al. (2020). Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies. PNAS, 117(32), 19061-19071.
6. Hatfield, E., & Walster, G. W. (1978). A New Look at Love. Addison-Wesley.
7. Gottman, J. M. (1999). The Marriage Clinic: A Scientifically Based Marital Therapy. W. W. Norton.