Repair Attempts: The 60-Second Move That Decides Whether a Fight Ruins the Week or the Marriage

You're 17 minutes into a fight that started about the dishwasher and has somehow become about your mother. Both of you have already said two things you mildly regret. The kitchen has that specific quality of air it gets when two people who love each other are also, right now, briefly, hating each other.

Then your partner says it, kind of awkwardly: "Wait. What are we even arguing about anymore?"

And there's a split-second. You can roll your eyes and answer with the next clean shot you had ready. Or you can let your shoulders drop half an inch and say "honestly, I'm not sure either." Whatever happens in the next ten seconds will largely determine whether tonight ends with both of you tired-but-okay on the couch, or with one of you sleeping in the guest room.

That ten-second window has a name. Couples psychologist John Gottman has been calling it a repair attempt for nearly forty years. And in his and his collaborator Robert Levenson's long-running observational studies of married couples, repair attempts are, more than almost any other variable he measured, the thing that distinguishes marriages that lasted from marriages that didn't.

The most surprising part of his finding isn't what most readers expect. It isn't that happy couples make repair attempts and unhappy couples don't. Almost everyone makes them. The differentiator is whether the partner on the receiving end notices, accepts, and lets the fight de-escalate. The sending is the easy half. The receiving is the half that decides the marriage.

What Repair Attempts Actually Are

A repair attempt is any small action or statement, made during or just after a disagreement, whose function is to step outside the fight for a moment and remind both partners that the relationship is bigger than the disagreement.

They can take almost any form:

• A small touch — a hand on the arm, a shoulder bump.
• A self-deprecating joke. ("Okay, I realise I just sounded exactly like my dad.")
• A meta-comment about the fight itself. ("Wait, what are we even arguing about?")
• An I-statement that names a feeling without blaming. ("I'm getting overwhelmed.")
• An explicit pause. ("Can we slow down for a sec?")
• A small acknowledgment of the other person's point. ("Okay, yeah, fair.")
• An apology for tone rather than substance. ("I didn't mean to sound that sharp.")
• A reference to a shared inside joke or memory that breaks the temperature.

What disqualifies an attempt isn't the form. It's whether it lands as a real offer of reconnection or as something sneakier — sarcasm, a manoeuvre to win the argument, a deflection. The same words can be both, depending on tone and what the partner already knows about the speaker.

One of the more useful things to understand about repair attempts is that they're often awkward. They almost have to be. By definition, they're someone choosing, in the middle of a tense exchange, to suddenly step out of the script. That kind of pivot rarely comes out smoothly. It often sounds slightly off, slightly forced, slightly out of place. That awkwardness is, weirdly, a feature. It's the audible sound of a person trying.

Where the Research Came From

Gottman, then at the University of Washington, started running couples through what would become widely known as the Love Lab in 1986, in long collaboration with Robert Levenson, who ran a parallel psychophysiology lab at UC Berkeley. Couples came in to discuss real ongoing conflicts while their physiological responses were measured in real time: heart rate, blood pressure, skin conductance, sometimes blood draws afterward. Every facial expression and verbal exchange was coded.

The original purpose was just to find out what conflict looked like, biologically and behaviourally, in stable versus distressed couples. What Gottman has reported in his books — most accessibly in his 1994 Why Marriages Succeed or Fail and his 1999 bestseller The Seven Principles for Making Marriage Work — is that one of the variables most strongly associated with which couples were still together at follow-up was the received rate of repair attempts during the recorded conflicts.

The often-quoted "Gottman can predict divorce with over 90% accuracy from a fight" claim that circulates online comes from this work. The honest framing is a little less dramatic: Gottman has reported high predictive accuracy in retrospective analyses of his observational data, using a model that combines repair attempts with several other variables including contempt, defensiveness, and physiological arousal. Independent replications have been more mixed, and the original methodology has had some pushback. But the core observation — that noticed and received repair attempts are one of the more reliable behavioural markers of long-term stability — has held up across his later work and into the broader couples-therapy literature, including Sue Johnson's Emotionally Focused Therapy framework, which builds repair into its central architecture.

Why Receiving Is the Harder Half

This is the part most readers find genuinely counterintuitive. If almost all couples send repair attempts, the question isn't how to send better ones. The question is why the same partner who can, calmly on a Sunday morning, recognise their partner's small attempts at reconnection completely fails to notice them on a Wednesday at 9 pm.

The cleanest answer is biological. During a high-intensity argument, the human nervous system enters what Gottman has called diffuse physiological arousal — heart rate climbing past roughly 100 beats per minute, stress hormones spiking, the prefrontal cortex partly going offline as the brain prioritises threat response over nuance. In that state, a partner's small "wait, can we slow down?" doesn't get processed as an offer of reconnection. It gets processed as a tactical move. Sarcasm. A trick. A manoeuvre to make you the bad guy. The receiver's brain is, in the moment, literally not equipped to receive it accurately. We've covered this body-level mechanism in much more depth in our piece on emotional flooding, which is the same phenomenon viewed from a different angle.

This is why repair attempts that get rejected aren't necessarily because the receiver is being mean. Often, the receiver genuinely isn't able to perceive what's being offered. Their body is in the wrong state for it. The most useful intervention isn't to try harder repair attempts. It's to interrupt the body state long enough that any repair attempt can be received.

How to Send a Repair Attempt That Actually Lands

None of these are clever. Most of them are slightly awkward. They work because the awkwardness is honest.

1. Lead with something small and concrete.

"Wait, what are we even arguing about?" lands better than abstract de-escalation language like "let's take a step back." The first one names what's happening. The second one sounds like a debate move. Specificity reads as honesty; abstraction reads as strategy.

2. Use the partner's name.

"Hey, Sam. Can we slow down?" lands differently than "can we slow down?" Names cut through the noise. They re-individualise the partner from "person I'm fighting" back to "person I love." It's a small linguistic trick. It works because no one is the abstract opponent your nervous system is currently treating them as.

3. Acknowledge something true the other person said.

This is the hardest one because it requires giving up ground while you're still defending. "Yeah, the dishwasher thing wasn't fair, you were right about that" is a real repair. It doesn't have to be the whole argument. One small concession often releases the tension enough that both of you can re-enter the actual disagreement with calmer bodies.

4. Be willing to be the first to break the script.

One of the harder lessons of long-term love is that someone usually has to go first, and that someone has to be willing to look slightly silly. The first person who softens during a fight risks being met with rejection. They do it anyway. Across the couples we've watched closely, the willingness to go first — repeatedly, even when it doesn't always work — is one of the more reliable markers of the partner who's playing the long game.

How to Receive a Repair Attempt When Your Body Is Yelling at You Not To

This is the half almost no one writes about, and the half the research actually identifies as the predictive one. Some practical guidance.

When Repair Doesn't Work

Not every repair attempt succeeds, and that's normal. The variable that matters is the pattern, not the individual instance. A relationship where repair attempts mostly land — even imperfectly, even slowly — is structurally different from a relationship where they mostly don't.

When repair consistently fails to land, the usual suspects are:

• The fights have built up too much heat. When physiological arousal is high enough, no repair can get through. The intervention isn't a better attempt; it's a real break. 20-30 minutes apart, no rehearsing the fight internally during that time, then come back.
• Contempt has crept in. Of Gottman's "Four Horsemen" of corrosive marital behaviour — criticism, contempt, defensiveness, and stonewalling — contempt is the one his research most consistently flags as the worst predictor. When contempt is present, repair attempts tend not to land because the receiver no longer trusts that they're real. The repair work in this case is much deeper and usually benefits from outside help.
• One partner has been doing all the repair. Repair attempts have to be roughly mutual over time. If you're always the one going first, and your partner consistently doesn't reciprocate, the imbalance itself becomes corrosive. The conversation that helps isn't "stop making repair attempts." It's making the asymmetry explicit and asking them to begin participating in the going-first.
• Quiet quitting has set in. If your partner has slowly disengaged from the relationship in the broader sense — the pattern we've covered in quiet quitting your marriage — repair attempts often fail to land because there isn't enough underlying investment for them to reach. The repair work has to start at the disengagement layer first.
• One of you is mentally somewhere else. The harder version of the above, and the one most people don't name out loud. When one partner is in limerence with someone outside the relationship — a real affair or, more commonly, a mental one — repair attempts at home stop landing because the partner offering the repair isn't quite who's being received. The work in this case isn't on the fight. It's on whether the relationship is actually still being inhabited by both people.

The Quieter Form: Pre-Conflict Repair

Most of the discussion of repair attempts focuses on what happens during a fight. But some of the most useful repair-shaped behaviour happens before any fight has started.

The texture of long-term love is mostly made of small everyday bids for connection — a comment, a question, a "look at this." We covered the in-person version in our piece on bird theory and the at-a-distance version in pebbling. Couples who turn toward those small bids reliably accumulate enough relational capital that, when a fight does break out, the repair attempts have something to land on. The bids and the repairs are made of the same material. The bids fill the bank when the relationship is calm. The repairs draw from it when the relationship is hot.

This is why couples in which the small daily turning-toward has been intact tend to recover from fights faster, even fights that look identical on the outside to those of couples without that bank account. The fight is the visible part. The accumulated bids are the invisible underwriter.

What This Means in Practice

You don't have to be eloquent. You don't have to be calm. You don't have to know what the fight is "really" about. You only have to be willing, at some point in the next exchange, to soften half an inch and offer something that doesn't quite fit the script of the argument. And — the harder half — when your partner does the same thing back, to notice it, accept it, and let the temperature drop, even if your body wants to keep going.

Over a year, a relationship that does this most of the time is doing the work that decides outcomes. Over a decade, it's the difference between two people who have grown together and two people who have, slowly and without intention, grown apart.

Frequently Asked

What's the difference between a repair attempt and just apologising?

An apology is one kind of repair attempt — and usually a high-impact one — but most repair attempts don't take the form of apologies. Apologies tend to be retrospective, after the harm has clearly been named. Repair attempts include all the smaller, in-flight gestures that try to interrupt the fight before it becomes the kind of fight someone needs to apologise for. A relationship doing well rarely needs many large apologies precisely because it has many small repairs.

Can humour really be a repair attempt?

Yes, and Gottman's research specifically identifies humour as one of the most effective repair vehicles in stable couples. The catch is that the humour has to be shared and self-deprecating rather than at the partner's expense. Sarcastic humour during a fight isn't a repair attempt; it's the opposite. A well-placed inside joke that lets both of you laugh at the absurdity of the situation is one of the cleaner repair moves available.

What if my partner makes repair attempts I don't believe are sincere?

That's information about a deeper layer of the relationship, not about the repair attempts themselves. If you've stopped trusting your partner's small de-escalations, the work is upstream of the next fight — it usually involves a longer, calmer conversation about what's been eroding the trust in general. Repeatedly rejecting in-flight repair attempts because the trust isn't there is rational in the short term and corrosive over years. If you find yourself there, this is one of the cases where couples therapy tends to be more useful than continuing to try alone.

How do you know if you're doing repair badly?

Two heuristics. First, ask whether your repair attempts tend to land with a clean response — even a small one — or whether they tend to be ignored or met with another shot. If they're consistently not landing, that's data about either the moment (your body is still too activated to make a real repair) or the pattern (something deeper is off). Second, ask whether your repair attempts include any actual concession of ground, or whether they're tactical pauses that allow you to re-launch the same point. Repairs require a small giving-up. Tactical pauses are a different move and usually felt as such.

Is it possible to use repair attempts manipulatively?

Yes, and it's one of the more painful failure modes. When repair attempts become a way to short-circuit a conversation the other partner needs to have — when "let's not fight" gets deployed to silence a legitimate concern — the form is being used to undermine the function. The signal is whether the partner using the repair language is actually willing to come back to the underlying issue once tempers have cooled, or whether the "let's not fight" was the whole point. Real repair attempts include an implicit commitment to return to the topic; tactical ones don't.

You don't need to be good at this. You only need to be willing to be the first to soften, often enough, that your partner can learn to receive softness without flinching. Most of what long love is made of is that exact, slightly awkward, mostly invisible exchange — happening many thousand times in the kitchen, in the car, mid-text, in the moment when no one else is looking. Couples who keep doing it usually stay together. Couples who stop usually don't. The mechanism is genuinely that small, and almost that simple.