You might assume couples who never fight are the ones who last. John Gottman’s decades of observation suggest otherwise: couples who stayed together had conflict too. The difference was not what they fought about, but how they fought.
It Is Not Conflict Itself—It Is This Kind of Conflict
Gottman found that the topic of a couple’s disagreement did little, on its own, to predict a relationship’s future. Couples can argue about money, parenting, or in-laws and still do well; other couples can break up over the very same issues.
The difference lies in the communication pattern. Couples who repeatedly criticize, show contempt, become defensive, or shut down during conflict—the “Four Horsemen”—are at substantially greater risk of relationship breakdown.
What Did Gottman’s Marriage Research Actually Study?
You may see the claim that Gottman tracked “3,600 couples for 39 years.” That is not a precise citation. Gottman has observed more than 3,000 couples across numerous studies over more than 40 years, beginning in 1972, but those figures do not describe one single study.
The best-known divorce-prediction study is Gottman and Levenson (1992), which followed 79 couples recruited in Bloomington, Indiana, for 14 years. After arriving at the lab following at least eight hours without talking to one another, couples discussed an area of conflict while researchers measured facial expressions, speech, heart rate, and other physiological responses.
Fourteen years later, 22 of the 79 couples—about 28%—had divorced. The widely repeated “3,600 couples over 39 years” figure therefore appears to combine and inflate numbers from different sources. The commonly cited 93.6% divorce-prediction figure comes from this line of research, but it also has limitations, discussed below.
The Four Horsemen: Warning Signs in How You Talk
Gottman identified four destructive patterns: criticism, contempt, defensiveness, and stonewalling. They can build on one another and undermine a relationship. The contrast is easiest to see in everyday language.
Criticism
- Instead of: “You are always like this. Why do you only think about yourself?”
- Try: “I felt hurt when you were late today. Could you let me know ahead of time next time?”
When a complaint about a behavior becomes an attack on someone’s whole character, the conversation stops being a discussion and becomes a trial.
Contempt
- Instead of: “That’s exactly why you’re such a mess,” with sarcasm, a smirk, or an eye roll.
- Try: “I’m getting angry about what you just said. I need a moment to collect myself, then let’s come back to this.”
Contempt includes mockery, cynicism, and dismissive expressions—any signal that treats your partner as beneath you.
Defensiveness
- Instead of: “What did I do wrong? You’re the one who is always late.”
- Try: “You’re right that I missed that. Can you tell me more about what was hard for you?”
Defensiveness often comes from feeling unfairly accused, but it can land as: “I am not going to hear you.”
Stonewalling
- Instead of: walking away without a word or staring at your phone.
- Try: “I’m too flooded to listen well right now. Can we take 20 minutes and come back to this?”
Stonewalling is often a sign of emotional overwhelm, not simple indifference. The key is to turn it into an explicit pause rather than silent withdrawal.
Why Contempt Is Especially Dangerous
Of the four patterns, Gottman identified contempt as the strongest predictor of divorce. Because contempt turns a partner from someone deserving respect into an object of ridicule, it can leave particularly deep injuries.
During an argument, a sarcastic tone, a demeaning nickname, a sigh, or an eye roll may look small but can be highly damaging signals. Continuing to respect your partner’s effort and basic worth is the relationship’s most basic line of defense.
How to Argue Better: A Gentle Start, Repair Attempts, and a Time-Out
Gottman does not ask couples to eliminate conflict. He recommends three skills for handling it.
First, use a gentle start-up. The opening minutes of a conversation often shape where it ends. Rather than starting with “Why do you always…,” begin with “I felt this way when….” That makes it more likely your partner can stay engaged instead of immediately getting defensive.
Second, make a repair attempt: a small move to reset the tone during conflict, such as a sincere apology, a light moment, or an offered hand. More durable relationships tend to be better at noticing and accepting these bids for repair.
Third, take a time-out when you are physiologically flooded. When anger spikes and your heart rate rises sharply, constructive conversation becomes genuinely difficult. Taking at least 20 minutes for an unrelated calming activity before returning can help. Saying, “We are not done talking about this,” is what distinguishes a time-out from stonewalling.
Be Honest About the Limits of Gottman’s Findings
Gottman’s work is widely cited, but it has also been criticized. Heyman and Slep (2001) cross-validated a prediction model by building it on one half of a sample and applying it to the other half. Accuracy dropped from roughly 90% in the training sample to roughly 69% in the validation sample.
That illustrates the risk of overfitting: a prediction formula developed after the fact for a particular group may not perform as well with a new group. “93.6% accurate at predicting divorce” is a result within the original research context, not a guaranteed probability for every couple.
Even so, the broader insight—that the way people communicate can matter more than the subject of a conflict—has continued to be supported in later research and clinical practice. Treat the headline numbers cautiously, but the communication insight remains highly practical.
Try This Today
In one line: breakups are shaped less by what couples fight about than by how they fight. In your next disagreement, try beginning your first sentence with “I” rather than “you.”
References
- Gottman, J. M., & Levenson, R. W. (1992). Marital processes predictive of later dissolution: Behavior, physiology, and health. Journal of Personality and Social Psychology.
- Heyman, R. E., & Slep, A. M. S. (2001). The hazards of predicting divorce without crossvalidation. Journal of Marriage and Family.