How do you know if a relationship can be saved?

EQTiger Editorial · Sources checked 2026-10-26 · Expert review pending

Why no test can give you the answer

If you have been taking online quizzes at midnight to find out whether your relationship is finished, here is why none of them can tell you. The famous claims that researchers can watch couples argue and predict which ones will divorce with startling accuracy came from equations fitted to the same couples they were then tested on. When an equation like that is checked against new couples, its accuracy falls sharply, which is why two researchers who reanalyzed archival data warned that uncross-validated divorce predictions deserve "extreme caution" (Heyman and Smith Slep, 2001). The patterns those studies describe are real enough to be worth knowing about, and what the four horsemen research actually shows sorts the solid parts from the oversold ones. They are not a verdict on you.

One of the largest attempts so far pooled 43 long-term couples studies from 29 labs, more than 11,000 couples in all, and ran them through machine learning (Joel and colleagues, 2020). How people felt about the relationship right now (whether they saw their partner as committed, felt appreciated, were sexually satisfied, fought a lot) explained up to 45 percent of the differences in how happy couples were at the start. Whether a relationship would get better or worse over the course of a study was, in the authors' words, "largely unpredictable" from anything people reported. If the best-funded version of this question cannot forecast change, a twelve-question quiz cannot either.

A question you can actually answer is a smaller one: do we both want to find out, and what would finding out look like?

What the breakup research does track

A meta-analysis of 137 studies and nearly 38,000 people asked which measures came before dating couples split up (Le and colleagues, 2010). The strongest signals were how committed each person was, how much they loved the other, how far they felt the partner had become part of who they were, and how much they depended on the relationship. Satisfaction and having attractive alternatives mattered only modestly. Friends' and family's support for the couple mattered a fair amount. Personality traits, including attachment style, predicted very little.

Two limits matter here. These were mostly unmarried dating couples, and the findings describe averages across thousands of people rather than odds for your particular relationship. Still, they point you at the right questions, and they are questions only the two of you can answer:

Answer them on your own first, in writing, on a calm day. Then notice whether your answers have been the same for months or swing with each argument.

Signs worth weighing, honestly

None of the lines below is a rule or a research finding. They are judgment calls, and each one is worth sitting with for a few days before you decide anything.

The same-argument-every-year line surprises people. Most couples have a few disagreements that never get solved, and having them is normal; how you talk about them is what changes things. Why some problems never get solved covers that in detail. An affair is not on the right-hand side by itself either. Plenty of couples recover from one, and rebuilding trust after cheating lays out what that work involves for each of you.

Here is one data point on how people end up at the courthouse. Among 886 divorcing parents surveyed in one Minnesota county in 2008, the two most common reasons for divorce were "growing apart" (55 percent) and "not able to talk together" (53 percent) (Hawkins, Willoughby, and Doherty, 2012). People who named growing apart, differences in tastes or money problems were less interested in reconciling. People who named abuse or adultery were neither more nor less interested. That is one county and one survey, but it suggests the quiet drift may be harder to reverse than the dramatic rupture, partly because by the time anyone names it, one partner has often stopped wanting to.

When one of you is already halfway out the door

Margot and Ewan have been married fourteen years. In March, on the drive home from his sister's wedding, Ewan said he did not know if he wanted to be married anymore. Margot wants to go to couples therapy tomorrow. Ewan does not want to "work on it." He wants to figure out whether there is anything left to work on. Every conversation since has gone the same way: she pushes, he goes quiet, and both of them feel more certain they are right.

That split is common. In a survey of 2,484 divorcing parents taking a required parenting class, about one in four said they believed their marriage could still be saved with hard work (Doherty, Willoughby, and Peterson, 2011). In about one in nine matched couples, both partners believed it. In about one in three matched couples, one partner was interested in reconciliation help and the other was not. The partner who had not filed was far more likely to be the interested one.

Standard couples therapy sets out to improve the relationship, which is hard to do when one partner has not decided they want it improved. Discernment counseling was designed for this mismatch: one partner leaning toward divorce, one wanting to keep the marriage (Doherty, Harris, and Wilde, 2016). The goal is not to fix the relationship. It is to help both people reach a clear, confident decision about what to do next, with most of the time spent in separate conversations with each partner. In an analysis of 100 consecutive cases, about half of the couples chose to start couples therapy to try to reconcile, and most of the rest chose divorce. That is one clinic's case series with no comparison group, so it does not prove the method works better than anything else. It does show that a leaning-out partner who agrees to look carefully does not always end up leaving.

If you are the one leaning in, this is how pushing tends to sound, and what works better:

You can't just give up on fourteen years. We're going to therapy, I already looked people up.

I don't want you to decide anything tonight. I'd like us to find someone who helps couples where one person isn't sure. Would you come to that?

If you are the one leaning out, this is the honest version of where you are:

I'm not asking for a divorce today. I'm saying I don't know, and I don't want to pretend I'm all in when I'm not.

A ninety-day trial you design together

If you both say yes to trying, treat it as a defined experiment rather than an open-ended maybe. Uncertainty with no end date wears people out.

1. Agree on the question (one evening, about an hour). Not "are we happy?" but something like: "Can we stop the weekly blowups about money and feel like a team again?"
2. Book a licensed couples therapist within two weeks. A review of a decade of couple therapy trials found therapy helped about 70 percent of the couples who received it (Lebow and colleagues, 2012). Trial couples are not identical to you, and no therapist can promise an outcome, but in those trials the couples who got no treatment did far worse.
3. Each of you names two specific changes you will make (fifteen minutes each). Your own, not your partner's. "I'll stop checking my phone at dinner" counts. "I'll be more present" does not.
4. Check in once a week for twenty minutes. Same day, same time, no rehashing of old fights. A simple weekly relationship check-in gives you a format.
5. At day ninety, each of you answers in writing: Is this better, the same, or worse? Do I want another ninety days?

The leaning-out partner is not agreeing to stay forever. They are agreeing to try for real until the review date, which means not keeping one foot out of the door the whole time. That is a fair thing to ask and a fair thing to offer.

When saving it is the wrong goal

Some relationships should not be saved, and the question itself can be a trap. If your partner threatens you, controls your money or phone, tracks where you go, pressures you into sex, or you are afraid of what happens when you disagree, that is not a communication problem, and more effort from you will not fix it. Joint counseling can also make it riskier to speak honestly when one partner fears the other. Talk to someone on your own first, and read hard situations and safety before you decide what to do next.

If the answer turns out to be no

Sometimes the ninety days end with both of you knowing it is over, or one of you knowing it clearly enough for both. That is a result, not a failure of the trial. If you are the one ending it, ending a relationship kindly covers the conversation itself, and if you share a home, breaking up with someone you live with covers the logistics. The rest of the pages on endings and recovery are in the breakups and moving on guide.

Start with the free EQTiger assessment (adults 18+), or go back to the Relationship Center.