← All questions

Relationships

a straight answer, then a way to find out

Why do I keep going back to my ex?

Because the thing that pulls you back — missing them, a bad week, a warm message — is never the thing that ended it, and nothing in the gap addressed the thing that ended it. On-again relationships are usually renewed on feeling and ended, again, on the same unresolved problem.

The cycle has a reputation as a failure of willpower, which is not what the research on on-again, off-again couples describes. It describes a structure: a relationship that is renewed for reasons that are real and never quite fixes the thing that broke it. The feeling that brings people back is genuine. It is simply not related to the problem, which is why the fourth reconciliation ends the same way as the first.

This page is about the loop. Why it feels the way it does is its own question, and the two are worth keeping separate — one is a sensation, this one is a mechanism.

What is usually going on

Renewal runs on feeling, endings run on problems. The decision to go back is typically made in a state — lonely, drunk, nostalgic, after a warm message at eleven p.m. The decision to end it was made about something structural: they would not move, they would not stop, the thing you needed was not on offer. A feeling can override a structure for about three weeks. The structure is still there in week four.

The break never worked on anything. Research on cyclical couples finds that partners who separate and reunite report more uncertainty about the relationship and more communication problems than couples who did not cycle — which makes sense, because time apart with no terms and no question attached is not a process, it is a pause. Nothing was tested, so nothing was learned, and the reunion starts exactly where the ending did.

Going back is cheaper than starting over. A known person, known routines, no first dates, no explaining yourself from the beginning. The cost of a familiar disappointment is genuinely lower than the cost of an unknown one, and that arithmetic is not stupid — it is just answering a different question from is this good.

The gap is unbearable and they are the fastest fix. The loneliness after an ending is worst at a predictable hour, and the person best equipped to end it in five minutes is the one who caused it. This is the version where the return is not about the relationship at all; it is about the evening.

Something genuinely did change. Occasionally it did. Someone stopped drinking, someone moved back, someone did the work. The tell is specific and boring: what changed is nameable, it changed before the reunion rather than being promised for after it, and it has held through something difficult.

Leaving is not currently safe. If the returns are driven by what they do when you go — the escalation, the threats, the crisis — then this is not a cycle question. That pattern is described in toxic relationships, and it needs people around you rather than a decision.

How to tell which return this is

Write down, before anything else, the reason it ended last time. Not the fight that ended it — the reason. Then ask three questions of the proposed reunion.

Has the specific thing changed, in behaviour, already? Not intention, not an apology, not a plan. A change that has been running long enough to have been tested. Promised change that starts on the day of the reunion has the worst record of anything in this area.

What is different about the version of me returning? If the answer is nothing, you are putting the same two people back into the same arrangement and expecting a different result from the arrangement.

Am I choosing this or ending an evening? The honest answer is usually available at three in the afternoon and almost never at midnight.

If all three come back clean, go back — plenty of relationships survive an ending and are better for it. If they do not, the useful move is not a stronger resolution. It is removing the mechanism: the late-hour contact, the open door, the ambiguous status that makes each return a small step rather than a decision.

What to do tonight

  • Write the ending reason down and keep it somewhere you will find it at eleven p.m. in three weeks. Your memory will have softened it by then; that is not a personal weakness, it is how memory works.
  • Count the cycles. Two is a story about two people. Four is a structure, and structures are changed by changing conditions rather than by deciding harder.
  • Make the late hour expensive. Almost every reconciliation in this pattern is made after ten p.m., and almost none of them would have been made at lunchtime.
  • If you are going back, go back to something different: one named change, already in place, with a date attached to when it started.
  • Tell one person who knew you during it. The cycle survives partly on privacy — most people stop mentioning the reunions somewhere around the third one, and the silence is what makes the next lap easy.

What it depends on

Cycles usually form around one specific mismatch rather than a general incompatibility — one person needs far more contact than the other, one reaches for the solution while the other needs the feeling heard first, one needs the plan settled before they can relax and the other finds a settled plan suffocating. The pull back is real because the parts that fitted genuinely fitted. The break is real because one dial never moved.

Naming the dial is what turns the fifth reunion into an actual decision: either the mismatch is negotiable, in which case there is something to work on, or it is not, in which case the cycle is the relationship and each return is just the next lap.

Which one are you? Find out in two minutes

The test reads four dials that decide day-to-day fit — closeness, how you read signals, whether you reach for the fix or the feeling, and how much certainty you need. A cycle usually forms around one specific mismatch, and naming which one turns a repeating story into a solvable question.

Want to talk it through?

Luna reads your results first, so the advice is about you, not about people in general. Five free messages a day.

✦ Talk to Luna

People also ask

How many times do couples break up before it's final?

Cycling is common enough that most people who study young adults' relationships treat it as an ordinary pattern rather than an unusual one. There is no threshold that makes it final — what predicts a lasting reunion is whether something specific changed, not how many attempts preceded it.

Do on-again, off-again relationships ever work?

Some do, and the ones that do tend to have a nameable change behind the final reunion — a move, a job, someone stopping something — rather than a renewed intention. Reunions built on missing each other have the weaker record.

Why do I go back when I know it won't work?

Because the knowing is stored in one part of the evening and the decision is made in another. Most returns happen late, alone, after contact. Changing the conditions works better than arguing with yourself about it.

Is it different if we're on a break rather than broken up?

Structurally, yes — a break with terms is a test, and an ambiguous one is a slow cycle with better branding. The distinction is worth getting right: [what a break actually is](/ask/is-a-break-in-a-relationship-a-good-idea).

Related questions

Written by MoodyWonder from the attachment, conflict, trust, occupational and sleep literature. Describes patterns; does not diagnose. If you are unsafe or everything has gone grey, talk to someone you trust or a clinician today.