The anxious–avoidant pairing has a reputation as the combination that ends in divorce. The reputation is half right, and the half that is wrong matters more: what damages the relationship is not the two styles sitting next to each other, it is one specific loop they fall into. The loop has a name, it is visible from the outside, and unlike a personality it can be interrupted.
What "anxious + avoidant" actually means
Attachment style describes what you do when closeness feels uncertain — not how much you love someone.
An anxious style responds to uncertainty by moving toward. More contact, more clarification, more attempts to fix it now. Psychologists call this a hyperactivating strategy: the system that scans for threat to the bond turns up, and it stays up until the threat is resolved.
An avoidant style responds to the same uncertainty by moving away. Fewer words, more space, a return to competence and independence. That is a deactivating strategy: the same threat detector turns down, because the lesson learned somewhere earlier was that needing someone does not reliably produce someone.
Both are answers to the same question — what do I do when I am not sure of you? Neither is the healthy one and neither is the broken one. They only become a problem in combination, and only because each person's answer is the other person's trigger.
If you have never established your own, the attachment style test is 25 questions, about eight minutes, and gives you the pattern rather than a label.
The mechanism: pursue and withdraw
Relationship researchers know this loop as the demand–withdraw pattern, and it runs the same way almost every time:
- Something small opens a gap. A short reply, a cancelled plan, a flat evening.
- The anxious partner moves toward it. Wants to talk, wants it resolved tonight.
- The avoidant partner feels the pressure to produce closeness on demand, and goes quiet — not to punish, but to regulate.
- Silence reads as confirmation of the original fear. The pursuit intensifies.
- The withdrawal deepens, because now the conversation is not about a cancelled plan, it is a referendum.
Nobody in this sequence is behaving badly. Each person is doing the thing that has always calmed them down, and each person's calming move is the other person's alarm.
That is why the fight is never really about the dishes or the phone. Two people in this loop are not arguing about the topic. They are arguing about whether closeness is safe, using the topic as the vehicle.
Why this pairing forms so often
If the combination is so difficult, it is fair to ask why it keeps happening. The answer is that early on, it feels unusually good.
Pursuit gives the avoidant partner proof of being wanted without having to ask for it — which is the exact thing they find hardest. Distance gives the anxious partner a familiar intensity: the emotional weather they grew up reading, and the sense that love is something you earn rather than something you are handed.
Two secure people falling for each other feels calm. Calm is not what the insecure system recognises as love. That is not a moral failing; it is a preference built by repetition, and it is why the early stage of an anxious–avoidant relationship often has more heat than anything either person has felt before.
What the research actually says — and what it does not
This section exists because the claim in the headline is repeated far more confidently online than the evidence supports.
What is well supported:
- Adult attachment is measurable and it predicts relationship behaviour. Hazan and Shaver's 1987 work carried Bowlby's infant categories into adult romance, and hundreds of studies since have found the styles are stable enough to be useful.
- The demand–withdraw pattern is one of the most reliably damaging things a couple can do. Christensen and Heavey's research on the pattern found it consistently associated with lower satisfaction and with decline over time — and the association holds regardless of who is doing which half.
- Withdrawal specifically forecasts trouble. Gottman's work on stonewalling — one partner shutting down mid-conflict — put it among the behaviours most corrosive to a marriage. Stonewalling is what deactivation looks like from the other side of the sofa.
What is not supported, and gets repeated anyway:
- That the anxious–avoidant pairing reliably ends in divorce. Kirkpatrick and Davis, following couples over several years, found the anxious-woman/avoidant-man pairing to be notably stable — while reporting low satisfaction. The combination may predict unhappiness far better than it predicts separation. Read honestly, that is a worse finding, not a reassuring one.
- That any study can tell you what will happen to your relationship. The widely quoted divorce-prediction accuracy figures come from models fitted to couples whose outcomes were already known. That is a description of a sample, not a forecast for a person.
- That a style is a verdict. Attachment styles move. They move with a different partner, with therapy, after a stable few years, and — the finding people forget — they move toward security more often than away from it.
The accurate version of the headline is this: the pairing does not cause divorce. The loop predicts misery, misery predicts divorce, and the loop is the part you can actually get hold of.
The four pairings, and how each one fails
| Pairing | How it goes wrong |
|---|---|
| Secure + anything | The stabiliser. A secure partner does not escalate and does not disappear, which starves both loops of fuel. This is the pairing where an insecure style most often shifts — one steady person is frequently enough to change the system. |
| Anxious + avoidant | Pursue–withdraw. Highest heat early, highest cost later. Each person's regulation strategy is the other person's threat, so the loop feeds itself without anyone intending it. |
| Anxious + anxious | Escalation with no brake. Nobody withdraws, so nothing de-escalates. Conflicts are long, intense, and end in exhausted reassurance rather than resolution. Often mistaken for passion. |
| Avoidant + avoidant | Parallel lives. Very little conflict, which both partners initially read as compatibility. It rarely explodes; it thins out, and one day there is nothing left to withdraw from. |
| Fearful-avoidant + anyone | Both moves inside one person. Pursuit and withdrawal alternate in the same partner, sometimes in the same conversation, which makes the pattern hardest to read from the outside — see fearful-avoidant attachment explained. |
Four things that actually change the odds
These are the interventions that survive contact with a real relationship, in rough order of leverage.
1. Name the loop instead of the person
"You are cold" and "you are needy" are both accusations, and both are wrong: each person is describing the other's coping move as if it were their character. "We are in the loop again" is the same observation with nobody cast as the villain, and it is the only version that leaves both people able to answer.
2. Withdrawal with a return time
The single highest-leverage change either partner can make, and it belongs to the avoidant one: say when you are coming back. "I need twenty minutes" is a different event from walking out of the room. The first is regulation. The second is, to an anxious nervous system, abandonment — and the pursuit that follows is a response to the missing return time, not to the space itself.
3. Contact that is scheduled rather than negotiated
When closeness has to be requested every time, the anxious partner spends the day deciding whether to ask and the avoidant partner spends it braced for the question. A fixed point in the week that neither person has to earn removes the negotiation entirely, and takes most of the daily pursuit with it.
4. Repair speed, not conflict avoidance
Couples who last are not the ones who argue least. They are the ones who come back fastest. If your loop currently takes three days to resolve, getting it to three hours changes the relationship more than eliminating the argument would — and it is a far more achievable target.
Which half of the loop is yours?
People are confident about their partner's style and wrong about their own. The fastest way past that is to stop asking how you feel and look at what you did.
Think about the last argument that did not resolve the same evening. In the first hour afterwards:
| In that first hour | Which half |
|---|---|
| You reread the messages, drafted three replies, checked whether they were online | The pursuing half |
| You found something to do, felt calmer once you were alone, and were surprised how upset they still were later | The withdrawing half |
| You did both, in that order, and the switch felt like a decision you did not make | Fearful-avoidant, the pattern that contains both |
| You were annoyed, said so, and then got on with the evening | Neither. This is what secure looks like, and it is less dramatic than people expect |
The point of the exercise is not the label. It is that whichever row you recognised, the move that helps is the opposite of the one your body wants.
If you are the anxious one
The move that helps is not "need less". It is to get the reassurance from more than one source, so that a single quiet evening is not carrying the entire weight of whether you are loved.
Two practical things, in order of difficulty.
Put a delay between the feeling and the message. Not to suppress it. Because the message you send after twenty minutes is almost always the one you would rather have sent.
Say the need directly instead of testing for it. "I would like an hour with you tonight" reliably outperforms a probing question about whether everything is fine. The probe is designed to find out whether you have to ask, and the answer it usually returns is the wrong one.
More on both in anxious attachment dating tips.
If you are the avoidant one
The move that helps is not "open up more". It is to make your withdrawal legible.
Most avoidant partners believe they are keeping the peace by going quiet, and are genuinely surprised to learn that silence lands as rejection. It does, reliably.
Announcing the space costs one sentence and defuses the entire second half of the loop: "I am not upset, I need an hour, I will come and find you." Three facts, none of which requires you to talk about the thing you do not want to talk about yet.
The avoidant attachment in relationships guide covers what deactivation feels like from the inside, and how to develop secure attachment covers the slower work.
When it is not attachment
Attachment theory explains a great deal, and it is also used to explain things it should not.
The loop described here is symmetrical: two people, both frightened, both making it worse for the other. That framing is only accurate when it is actually symmetrical.
- Contempt is not avoidance. Someone who mocks you, in private or in front of others, is not regulating their nervous system.
- Control is not anxiety. Monitoring your phone, your money or your friendships is a different thing from needing reassurance, and it does not respond to reassurance.
- One-sided effort is not a pairing. If every repair, every scheduled evening and every naming of the loop is yours, you are not in a pattern with someone. You are maintaining a relationship alone.
Calling any of those an attachment style keeps a person working on their own responses while the real problem goes unnamed. If the situation involves fear rather than frustration, attachment is the wrong lens entirely.
Where to start
Work out which half of the loop is yours before deciding what to do about the relationship. Most people are confident about their partner's style and wrong about their own — the anxious partner tends to read their pursuit as reasonable, and the avoidant partner tends to read their withdrawal as neutral.
The free attachment style test gives you your own pattern in about eight minutes, including the fearful-avoidant profile that most short quizzes miss. If you want the theory behind it first, start with the complete guide to attachment theory.
One last thing worth holding on to. Of everything in this article, the finding with the best evidence behind it is also the most useful: security is contagious. One partner who stops escalating and stops disappearing changes the loop for both people — and does not need the other one's permission to start.