Guides · Conflict and repair
The four horsemen of relationships: how to spot them, and what to do instead
Criticism, contempt, defensiveness and stonewalling: the four ways of arguing that John Gottman’s research flagged as trouble. Here’s how to spot each one (in yourself, too), what to do instead, and what the studies can and can’t tell you.
By The Couples Company15 min read20 sources

The short version
- The four horsemen are criticism, contempt, defensiveness and stonewalling, four ways of arguing that John Gottman’s studies tied to unhappier marriages and divorce.
- Complaints aren’t the problem: a horseman turns a complaint into an attack on who your partner is, or shuts the conversation down.
- Stonewalling often comes from feeling flooded, and a break is easier to trust when you say when you’ll be back, then come back.
- When your partner is the one doing it, answer the behavior without a horseman of your own, and talk about the pattern when you’re both calm.
- The famous divorce-prediction figures come from small studies that weren’t tested on new couples, so treat the horsemen as warning signs, not a forecast for your own relationship.
Not a diagnosis. This guide explains patterns that researchers study. It isn’t a diagnosis or a substitute for therapy. If these patterns weigh on you or your relationship, a licensed therapist can help.
What are the four horsemen?
The four horsemen are criticism, contempt, defensiveness and stonewalling: four ways of arguing that psychologist John Gottman found were more closely tied to separation and divorce than other kinds of negativity in his early studies of married couples. He named them after the four horsemen of the Apocalypse.1
In the framework the Gottman Institute teaches, each one has a recognizable shape:2
- Criticism: turning a complaint about something your partner did into a verdict on who they are. “You never think ahead.”
- Contempt: treating your partner as beneath you, with mocking, sneering, eye-rolling, name-calling or sarcasm that’s meant to sting.
- Defensiveness: protecting yourself with excuses, counterattacks or wounded innocence, so the problem lands back on your partner.
- Stonewalling: going blank while your partner is talking: no eye contact, a still face, no sounds that say you’re following.3
In Gottman’s early data, a statistical model fit a sequence in which criticism leads to contempt, contempt to defensiveness, and defensiveness to stonewalling.1 Notice what isn’t on the list: anger. In his studies, anger on its own didn’t predict divorce.1,4
And before you start keeping score: these show up in happy couples too. In a study of 130 newlywed couples, answering negativity with negativity was common even in happy, stable marriages.4 In a 1993 study of 73 couples, even stonewalling looked harmful mainly when it was frequent or wasn’t balanced by warmth: the groups of couples that proved more stable over four years showed about five positive behaviors for every negative one while they argued, and the groups more prone to divorce about one to one, or worse.3
What do the four horsemen look like in real life?
In real life, the four horsemen sound like ordinary arguing, and each has a healthy cousin that looks a lot like it, which is why they’re easy to miss, especially in yourself.
Criticism. A complaint is about a moment: “You said you’d call the plumber, and it didn’t happen.” Criticism is about a person: “You never follow through on anything.” Watch for “always,” “never” and “What kind of person…” Complaints themselves are healthy, so say the thing. Just aim it at the behavior. And don’t sit on a serious problem to keep the peace: in two studies of newlyweds, being direct, even blaming, about a serious problem was linked to the problem improving and satisfaction holding steady, while the same directness about minor issues went with steeper declines.5
Contempt. This is criticism with a sneer: an eye-roll mid-sentence, mimicking their voice, “Oh, here we go,” name-calling. It says “I’m above you.” The Gottman Institute calls contempt the greatest predictor of divorce.2 In the data behind Gottman’s theory, contempt was linked to divorce for husbands and wives alike (though not most strongly, as you’ll see),1 and in his newlywed study, contempt, belligerence (goading or daring your partner) and defensiveness together predicted divorce.4 A team at UCLA found a similar warning sign: among 136 newlywed couples who stayed happy through their first four years, the ones who had divorced by year ten had shown more anger and contempt when they discussed problems as newlyweds.6
Defensiveness. It’s the reflex to protect yourself instead of hearing the complaint: an excuse (“I was going to, but work…”), a counterattack (“Well, you forgot the car insurance”) or the wounded “So I can’t do anything right?” It feels like self-defense. To your partner, it sounds like the blame just bounced back. In the data behind the original theory, defensiveness was actually the horseman most strongly tied to divorce, for husbands and wives alike (for wives, only just ahead of contempt),1 and when those couples were followed for 14 years, it was one of the clearest markers of the ones who divorced early.7
Stonewalling. The blank face, the eyes drifting to a phone, the one-word answers, or walking out with no word on when you’ll be back. In the Gottman framework, it usually happens when someone feels flooded, and it gets its own section below.2
| The horseman | Its healthy cousin | How to tell them apart |
|---|---|---|
| Criticism: “You’re so inconsiderate.” | A complaint: “I was hurt that you made Saturday plans without asking me.” | A complaint names one thing that happened. Criticism names a flaw. |
| Contempt: a sigh, an eye-roll and “Wow, genius.” | Frustration, said straight: “I’m really annoyed right now.” | Frustration is about how you feel. Contempt is about how little you think of them. |
| Defensiveness: “Me? You’re the one who never answers texts.” | Explaining: “I forgot. Work ate my week. I’ll do it tonight.” | An explanation still owns the slip. Defensiveness hands it back. |
| Stonewalling: silence and a blank stare | A break: “I’m too wound up to listen well. Back in half an hour.” | A break comes with a reason and a return time. A wall comes with neither. |
What are the antidotes to the four horsemen?
The Gottman Institute pairs each horseman with what it calls an antidote: a gentle start-up for criticism, a culture of appreciation and respect for contempt, taking responsibility for defensiveness, and physiological self-soothing for stonewalling.2 Here’s how each one can look in an ordinary week, in our words.
- 1
Start gently, not with a charge sheet
Open with the situation and what you’d like, not a verdict. “The recycling’s overflowing again. Could Tuesdays be your night?” lands very differently from “Do you even notice anything that needs doing?” Openings are telling: in a study of 124 newlywed couples, the 17 who divorced within six years had already been more negative and less positive in the first three minutes of a conflict discussion than the couples who stayed married.8
- 2
Make appreciation a habit
A steady habit of thank-yous leaves less room for sneering. Notice the coffee, the paid bill, the call they made, and say so. In the 1993 study, the more stable marriages weren’t free of negativity; they had enough warmth to outweigh it,3 and in the newlywed study, positive moments during conflict (things like humor, affection and interest) were the only thing that predicted both staying together and being happy.4 Our guide to emotional intimacy has small daily ways to build that warmth.
- 3
Own your part, even a small one
Find the piece of the complaint that’s true and say it first. “You’re right, I said I’d handle it, and I didn’t. I’ll do it tonight.” You don’t have to agree with everything to agree with something. If your apologies tend to miss each other, our apology quiz shows what helps each of you make up after a fight.
- 4
Take a real break
When you’re too flooded to listen, say so, step away, calm your body down and come back. That last part is what makes it a break, and the next section covers how.
- Can I bring something up? It’s small, but it’s been bugging me.
- I don’t think I ever said thanks for sorting out the car. That was a big help.
- You’re right about that part. I dropped the ball.
- Okay, that came out wrong. Can I try again?
- I’m getting too heated to be fair. Can we pick this up at eight?
What is stonewalling, and how do you stop it?
Stonewalling is checking out of a conflict while it’s still going, and you stop it by trading the wall for a real break, with a time to come back. The stonewaller stops answering, stops looking at you and stops showing they’re following. Gottman described it as the listener withdrawing from the conversation.1,3 From the outside it looks like not caring. From the inside, it often feels like too much.
Gottman called that too-much feeling flooding: your partner’s negativity seems sudden and uncalled for, it feels intense and overwhelming, it scrambles your thinking, and all you want is out.1 In a study of 233 couples, people who said they flood more easily tended to have partners who showed more anger, and couples with at least one partner prone to flooding solved problems less effectively.9 Your body is part of it. In a small early study that followed up 19 couples three years later, the more physiologically aroused couples had been during their conversations, the more their satisfaction had declined.10
Is stonewalling a guy thing? In Gottman’s early studies of married couples, husbands did more of it.1,7 But in the 233-couple study, men and women reported almost exactly the same amount of flooding,9 and who pulls back depends a lot on whose issue is on the table. In a study of 31 married couples, each spouse pushed harder for a change they wanted and pulled back more when their partner wanted one, though husbands withdrew more overall.11 In a study of 63 gay, lesbian and straight couples, whoever wanted more change did more of the pushing, whatever the couple’s genders.12
The fix isn’t to never step away. It’s to step away in a way your partner can trust. In a study that followed 373 newlywed couples for 16 years, couples in which one partner tried to talk things through while the other went off to cool down had higher divorce rates.13 So make your breaks easy to trust:
- 1
Agree on a signal now
Pick a word or phrase that means “I’m flooding and need a pause,” and agree that either of you can use it. Agreed on in a calm moment, it’s a shared tool, not a slammed door.
- 2
Say why, and say when
“I’m too wound up to listen well. Can we pick this up at eight?” The return time is what turns walking away into a break.
- 3
Calm your body, not your case
Walk, shower, breathe slowly, put on music. Gottman’s original advice was to do something soothing during a break and not to replay the argument in your head.1 He also suggested taking one when your heart rate climbs 10 or more beats a minute above its resting rate, as part of a therapy he proposed rather than tested.1
- 4
Give it real time
The Gottman Institute suggests at least 20 minutes.2 We couldn’t find a study that tested how long a real-life break should be, but even very short pauses have helped in the lab: in an experiment with 52 young couples, mostly students, the winner of each round of a game picked how loud a noise to blast at the other, and pauses of just 5 to 15 seconds before choosing lowered the blasts when people were visibly upset.14
- 5
Come back when you said you would
Even if it’s only to say you need until tomorrow. Open with something that lowers the temperature: “Thanks for waiting. I want to hear the rest.”
| Compare | A break | Stonewalling | The silent treatment |
|---|---|---|---|
| What’s going on | You’re calming down so you can talk well | You’re overwhelmed and shutting down | You’re freezing your partner out to make a point |
| How long | Minutes or hours, with a time to come back | Usually the rest of the argument | Hours or days |
| What your partner hears | “I’ll be back.” | “I’m not here.” | “You don’t deserve an answer.” |
| What helps | Coming back when you said you would | Noticing the flood and calling a real break | Saying what’s wrong, even if it’s “I’m hurt and not ready to talk yet” |
What do you do when your partner is the one doing it?
Answer the behavior without reaching for a horseman of your own, and save the bigger talk about the pattern for a calm moment. Meeting criticism with defensiveness, or a wall with louder knocking, is how one bad moment turns into a whole bad evening.
- If they criticize: translate it into the complaint underneath and answer that. “Sounds like you’re fed up with doing the dishes alone. Is that it?”
- If they get defensive: go softer, not louder. A short review of conflict research suggests warmth and validation can work better when a partner is defensive or the problem is minor, while a serious problem that needs to change still calls for being direct.15
- If they stonewall: don’t chase. Name what you see and offer the pause: “You look done. Want to stop and pick it up after dinner?” Following them from room to room won’t calm either of you down.
- If they show contempt: name it calmly and set a limit. “I want to talk about this, but not while you’re mocking me.” Then follow through. You’re allowed to ask for respect, and our guide to boundaries in relationships can help you find the words.
Listen for the repair under the tone, too. Gottman noticed that people often try to patch up a fight in a less than gentle voice, and that couples in distress tend to react to the snap, while happier couples respond to the request.1 “Can you let me finish?” through gritted teeth is still a request you can grant.
If it’s always the same dance, one of you pushing and the other shutting down, you’re in a well-studied pattern. Across 74 studies with more than 14,000 people, this push and pull (researchers call it demand-withdraw) had a moderate link to less satisfying relationships and poorer communication, whichever partner did the pushing.16 Our guides to anxious attachment and avoidant attachment dig into why some of us chase and some of us retreat, and the conflict style quiz shows where each of you lands on talking now or cooling off first.
When contempt or stonewalling has been around for a long time, trust can take a hit too, and our guide to rebuilding trust is a good next read. If you keep landing in the same fight, a couples counselor can help you both see the pattern from the outside. That’s a smart move, not a last resort.
Can the four horsemen really predict divorce?
Not for any one couple. The four behaviors are well studied, and they’re linked to unhappier marriages and to divorce.1,7,17 But the famous line that they predict divorce with more than 90 percent accuracy needs a big asterisk.
Here’s where numbers like that come from. In a 14-year study of 79 couples, 22 of whom divorced, the paper’s summary says a model combining the couples’ satisfaction, their thoughts of divorce and the emotions they showed in two conversations predicted divorce with 93 percent accuracy. The results themselves report separate models for early and later divorces, which sorted the couples who divorced from the rest correctly 75 to 88.5 percent of the time.7 In the newlywed study, 17 of 130 couples divorced within three to six years, and when those 17 were compared with the 20 happiest and 20 unhappiest couples who stayed together, the prediction was right 83 percent of the time.4 Every one of these models was built and scored on the same couples, and neither paper reports testing them on a fresh group.4,7
When researchers took older data and checked prediction formulas on couples they weren’t built from (cross-validation), accuracy and usefulness dropped steeply. They warned that results without that check should be read with extreme caution, and that predictions also have to account for how common divorce is in the first place.18 Another team argued that the advice drawn from the newlywed study went beyond what its data could show, citing how the couples were chosen and strong cause-and-effect conclusions drawn from correlations.19 Gottman himself called his theory highly speculative when he first laid it out, noting that it rested on correlational data from a few short-term studies.1
None of this means the horsemen are made up. Other labs have found related warning signs. In a 16-year study of 373 newlywed couples, spouses who said they yelled, insulted or called their partner names, dredged up old history or had to have the last word had higher divorce rates,13 and the UCLA study above linked anger and contempt to later divorce even in happy newlyweds.6 The arrow also runs both ways: in a study of 431 low-income newlywed couples observed four times over their first three years, how satisfied couples were predicted their later communication at least as well as their communication predicted later satisfaction, and neither link was especially strong.20
So treat the horsemen as a mirror, not a crystal ball. They describe habits that, on average, go with harder marriages, and habits can change. The useful question isn’t “Are we doomed?” but “Which one do we each reach for, and what will we try instead next time?”
Questions people ask
Which of the four horsemen is the worst?
Is stonewalling the same as the silent treatment?
Is stonewalling a form of abuse?
Can a relationship recover from the four horsemen?
Do the four horsemen apply to dating or same-sex couples?
Is this affiliated with the Gottman Institute?
Sources
Every number and finding in this guide comes from these sources, so you can check our work.
- 1.Gottman, J. M. (1993). A theory of marital dissolution and stability. Journal of Family Psychology, 7(1), 57 to 75. doi.org/10.1037/0893-3200.7.1.57Review
- 2.The Gottman Institute. (2013, April 26; updated 2026, October 29). The Four Horsemen: The antidotes. The Gottman Institute blog. gottman.com/blog/the-four-horsemen-the-antidotesWeb page
- 3.Gottman, J. M. (1993). The roles of conflict engagement, escalation, and avoidance in marital interaction: A longitudinal view of five types of couples. Journal of Consulting and Clinical Psychology, 61(1), 6 to 15. doi.org/10.1037/0022-006X.61.1.6Study
- 4.Gottman, J. M., Coan, J., Carrère, S., & Swanson, C. (1998). Predicting marital happiness and stability from newlywed interactions. Journal of Marriage and the Family, 60(1), 5 to 22. doi.org/10.2307/353438Study
- 5.McNulty, J. K., & Russell, V. M. (2010). When “negative” behaviors are positive: A contextual analysis of the long-term effects of problem-solving behaviors on changes in relationship satisfaction. Journal of Personality and Social Psychology, 98(4), 587 to 604. doi.org/10.1037/a0017479Study
- 6.Lavner, J. A., & Bradbury, T. N. (2012). Why do even satisfied newlyweds eventually go on to divorce? Journal of Family Psychology, 26(1), 1 to 10. doi.org/10.1037/a0025966Study
- 7.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 the Family, 62(3), 737 to 745. doi.org/10.1111/j.1741-3737.2000.00737.xStudy
- 8.Carrère, S., & Gottman, J. M. (1999). Predicting divorce among newlyweds from the first three minutes of a marital conflict discussion. Family Process, 38(3), 293 to 301. doi.org/10.1111/j.1545-5300.1999.00293.xStudy
- 9.Malik, J., Heyman, R. E., & Smith Slep, A. M. (2020). Emotional flooding in response to negative affect in couple conflicts: Individual differences and correlates. Journal of Family Psychology, 34(2), 145 to 154. doi.org/10.1037/fam0000584Study
- 10.Levenson, R. W., & Gottman, J. M. (1985). Physiological and affective predictors of change in relationship satisfaction. Journal of Personality and Social Psychology, 49(1), 85 to 94. doi.org/10.1037/0022-3514.49.1.85Study
- 11.Christensen, A., & Heavey, C. L. (1990). Gender and social structure in the demand/withdraw pattern of marital conflict. Journal of Personality and Social Psychology, 59(1), 73 to 81. doi.org/10.1037/0022-3514.59.1.73Study
- 12.Holley, S. R., Sturm, V. E., & Levenson, R. W. (2010). Exploring the basis for gender differences in the demand-withdraw pattern. Journal of Homosexuality, 57(5), 666 to 684. doi.org/10.1080/00918361003712145Study
- 13.Birditt, K. S., Brown, E., Orbuch, T. L., & McIlvane, J. M. (2010). Marital conflict behaviors and implications for divorce over 16 years. Journal of Marriage and Family, 72(5), 1188 to 1204. doi.org/10.1111/j.1741-3737.2010.00758.xStudy
- 14.McCurry, A. G., May, R. C., & Donaldson, D. I. (2024). Both partners’ negative emotion drives aggression during couples’ conflict. Communications Psychology, 2(1), article 73. doi.org/10.1038/s44271-024-00122-4Study
- 15.Overall, N. C., & McNulty, J. K. (2017). What type of communication during conflict is beneficial for intimate relationships? Current Opinion in Psychology, 13, 1 to 5. doi.org/10.1016/j.copsyc.2016.03.002Review
- 16.Schrodt, P., Witt, P. L., & Shimkowski, J. R. (2014). A meta-analytical review of the demand/withdraw pattern of interaction and its associations with individual, relational, and communicative outcomes. Communication Monographs, 81(1), 28 to 58. doi.org/10.1080/03637751.2013.813632Review
- 17.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 to 233. doi.org/10.1037/0022-3514.63.2.221Study
- 18.Heyman, R. E., & Smith Slep, A. M. (2001). The hazards of predicting divorce without crossvalidation. Journal of Marriage and Family, 63(2), 473 to 479. doi.org/10.1111/j.1741-3737.2001.00473.xStudy
- 19.Stanley, S. M., Bradbury, T. N., & Markman, H. J. (2000). Structural flaws in the bridge from basic research on marriage to interventions for couples. Journal of Marriage and the Family, 62(1), 256 to 264. doi.org/10.1111/j.1741-3737.2000.00256.xReview
- 20.Lavner, J. A., Karney, B. R., & Bradbury, T. N. (2016). Does couples’ communication predict marital satisfaction, or does marital satisfaction predict communication? Journal of Marriage and Family, 78(3), 680 to 694. doi.org/10.1111/jomf.12301Study
How we write our guides. From published research, in plain English, with every source linked. Spotted something we got wrong? Tell us.


