The Hidden Costs of Staying in an Unhappy Marriage

The costs of divorce are loud. The financial disruption, the custody logistics, the social rearrangement — they announce themselves on arrival and show up in every article about separation. The costs of staying in a chronically distressed marriage are quieter. They are paid in cortisol, in sleep, in immunological markers, in a slow erosion of energy that can look, from the outside, like getting older. Four decades of research on marital distress have been, in effect, about measuring the quiet costs. This article is not an argument for leaving. It is an argument that the second column of the ledger exists.

The default assumption

When people weigh divorce, the staying side of the decision usually shows up as a zero. Leaving has costs; staying is treated as the status quo — not the no-loss option, exactly, but the option with no new losses on top of what is already there. This is the baseline most of us bring to the problem, and it is the baseline most of the research in this article contradicts.

The staying side is not a zero. It is a column of costs that accrue slowly and are easy to miss precisely because they have been accruing for years. Making them visible is not the same as making a recommendation; it is an attempt to level the ledger so the decision is made with both columns filled in.

The physiological column

The most rigorous work on the biology of marital distress comes from Janice Kiecolt-Glaser and Ronald Glaser's lab at Ohio State. Across a long series of studies, they brought married couples into a controlled setting, had them discuss a point of real disagreement, and measured physiological markers before and after. The findings were consistent and striking. Hostile marital interactions produced elevated stress hormones, elevated pro-inflammatory cytokines, and — in a widely replicated 2005 study — measurably slower healing of small experimental wounds. The mechanism is not mysterious: sustained low-grade hostility keeps the stress axis activated, and chronic activation has downstream effects on inflammation and immunity.

A decade of follow-on work extended the finding across disease endpoints. Robles and colleagues' 2014 meta-analysis in Psychological Bulletin pooled the published literature and reported that marital quality was consistently associated with health outcomes across multiple domains — cardiovascular, immunological, endocrine — and that the effect was stronger than many earlier single studies had suggested. Holt-Lunstad's work showed that, on some markers like ambulatory blood pressure, being in a low-quality marriage was worse than being single.

The point is not that an unhappy marriage is a medical emergency. It is that the costs are real and physiologically traceable, not a matter of mood. Many people carry these costs for years without naming them.

The children column

"Staying for the kids" is one of the most common reasons people give for remaining in a distressed marriage, and the intention is legitimate. The research on outcomes for children, however, complicates the default version of the argument.

Mavis Hetherington's Virginia Longitudinal Study tracked children across three decades into adulthood. She reported what has become the most important finding in the post-divorce developmental literature: ongoing parental conflict — not the legal event of divorce — was the strongest predictor of negative outcomes in children. Children in low-conflict divorced families often did better than children in high-conflict intact families. Paul Amato and Alan Booth's longitudinal work converged on the same conclusion.

The relevant comparison for parents considering divorce is therefore not "divorce versus happy marriage." It is "divorce versus the marriage the children actually live inside." If the marriage is low-conflict and simply unsatisfying, staying probably is the better environment for the children. If the marriage is high-conflict and chronic, the developmental evidence does not support staying as the protective option — staying is the exposure option.

The self column

Less quantifiable but not less real are the costs to the person doing the staying. David Sbarra's research on divorce and health sits alongside the Kiecolt-Glaser and Robles work; it also examines what happens psychologically when attachment systems cannot complete their natural cycles. Chronic distressed bonding — being tied to a person with whom the co-regulation loop is broken — produces a state Sbarra describes as dysregulation: the attachment system is activated but not resolved. Over years, people describe this state in recognizable ways. A lowering of baseline energy. A dulling of emotional range. A sense, not easily named, of carrying something heavy.

Some of this is reversible. Sbarra's post-divorce work finds that many people recover psychological range within one to two years of separation, and that the recovery is most reliable when the marriage had been chronically distressed before the split. This is a form of evidence the staying calculation rarely includes — not as an argument for leaving, but as a correction to the assumption that the current baseline is permanent and costless.

What this article is not

Several things follow from the research above, and several things do not. It may help to be explicit.

• This is not an argument that leaving is always better. Waite et al.'s 2002 finding — that most unhappily married adults who stayed were happy again five years later — still holds at the population level. The physiological and developmental costs discussed above concentrate in the sub-population of high-distress marriages, not in all unhappy marriages.
• This is not an argument that children are better off divorced. Hetherington's finding is specifically that high-conflict intact families are the worst developmental environment — worse than divorce. Low-conflict marriages remain, on the evidence, better environments than divorce for most children.
• This is not an argument to make a decision quickly. The same cautions about premature resolution in the ambivalence article apply here. The purpose of visibility is better decisions, not faster ones.

What the research does support is a more honest ledger. The costs of staying are not zero. Putting them on the page alongside the costs of leaving does not decide the question, but it changes which question you are answering.

A useful self-check

If you want to take the research above and do something small with it, consider noticing — over the next month — the specific markers the studies track. Not to diagnose yourself, but because attention is a form of data.

• Sleep. Research on marital distress consistently finds disrupted sleep as an early indicator, particularly for the partner doing more of the emotional work.
• Body. Notice the tightening in your shoulders, jaw, or chest during specific recurring interactions. The body tends to learn which conversations are dangerous before the mind agrees.
• Affect. The flatness Sbarra describes often appears as a narrowed range — less delight, less irritation, less of either direction. When people describe feeling "numb," this is often what they are describing.
• Presence with the kids. If your children's faces make you brighter in a way your partner's no longer does, this is a piece of information. It does not mean the marriage is over; it does mean something has shifted.

None of these is a verdict. They are markers the research treats as informative, and they are usually available to the person carrying them long before they become large enough to be impossible to ignore.

The bottom line

The cultural script treats divorce as the decision with consequences and staying as the decision with none. Four decades of research on marital distress suggest that is not the right frame. Both decisions have consequences. What varies — across individuals, across children, across years of runway — is which column is heavier, and that is the question any honest decision has to engage.

Common questions