How long does the average relationship last?

The honeymoon effect refers to the period of elevated positive sentiment at the start of a relationship, characterised by heightened attraction, optimism, and idealisation of a partner. Research documents that this phase typically lasts 12 to 30 months and is associated with neurobiological changes including elevated dopamine and reduced serotonin. The transition out of the honeymoon phase is a normal developmental stage, not a sign of relationship failure. Couples who build genuine friendship and shared meaning during this period show better long-term outcomes than those who do not.

Research by Rhoades and Stanley at the University of Denver, tracking 1,600 Americans from 2010 to 2019, found that pre-engagement cohabitation was associated with a 34% divorce rate, compared to 23% for couples who cohabited only after becoming engaged. The difference is attributed to the "sliding versus deciding" mechanism: couples who move in together before committing may do so for logistical convenience rather than relationship commitment, and may subsequently feel locked into marriage without a deliberate decision having been made.

Researchers have identified a pattern called the terminal transition point, a period of rapid, largely irreversible deterioration that occurs 1 to 2 years before relationship dissolution. This phase, lasting approximately 7 to 28 months, is characterised by accelerating negative reciprocity, reduced repair attempts, and progressive withdrawal. It differs from ordinary conflict cycles in that recovery rates are substantially lower. Identifying and intervening before this transition is one of the central goals of couples therapy research.

Earlier research from Cacioppo et al. (2013) suggested online-met couples had a slight longevity advantage. However, a 2024 study by Kowal and colleagues across 50 countries with 6,646 participants found that offline-met couples now report higher satisfaction, passion, commitment, and lower break-up rates. The current balance of evidence favours offline-met couples on longevity metrics, though the effect size is moderate and many online-met couples form highly stable relationships.

General Social Survey data and analyses drawn from the NCFMR suggest that cohabiting relationships that ultimately dissolve have a median duration of approximately 1.3 years. This is substantially shorter than the median for cohabiting relationships that transition to marriage, which tend to last several years before the decision to marry is made. The high dissolution rate of short-term cohabiting relationships is one reason why population-level statistics on cohabitation show lower stability than marriage statistics: the pool includes many short experimental partnerships that never intended permanence alongside longer committed cohabiting relationships.

The popular concept of a seven-year crisis point in marriage is not strongly supported by the research evidence. CDC divorce timing data and analyses of US marriage cohorts find that dissolution risk is highest in the early years of marriage, particularly years two through five, and declines steadily after that. Some studies have identified a secondary elevation in risk around years 10 to 15, which may account for the cultural salience of the 7-year figure as a rough average. Gottman's longitudinal work on couple dynamics did not find a year-specific crisis point: instead, he identified specific interaction patterns (contempt, stonewalling, criticism, defensiveness) as predictors, regardless of relationship duration. The 7-year itch is probably best understood as a cultural narrative rather than a reliable statistical phenomenon.

John Gottman's research at the University of Washington, tracking couples over decades, identified the ratio of positive to negative interactions during conflict as the single strongest predictor of relationship stability. Stable couples maintained a ratio of approximately five positive interactions (affirmations, humour, physical affection) for every one negative interaction even during disagreements. Beyond this, research consistently identifies effective conflict resolution, shared meaning and life goals, friendship and genuine liking for the partner, and secure attachment styles as the strongest predictors of longevity. Age at partnership also matters: couples who partner in their mid-to-late twenties show better average longevity outcomes than those who partner in their teens, partly because identity and values are more stable by then.

US Census Bureau data finds that the median time between divorce and remarriage is approximately 3.5 years for those who do remarry. Roughly 64% of divorced Americans remarry at some point, though this rate has declined over recent decades as long-term cohabitation has become a more accepted alternative. Men tend to remarry faster than women on average, with a median of around 3.3 years compared to 3.7 years for women. Younger divorcees remarry faster than older ones. Second marriages have higher dissolution rates than first marriages, with some studies finding approximately 67% of second marriages ending in divorce, compared to around 41% for first marriages.

Strongly. CDC data and sociological research consistently find that marriages formed in the teen years have markedly higher dissolution rates than those formed in the twenties or early thirties. The divorce rate for marriages formed at age 18 is approximately double that for marriages formed at age 25 or later. This relationship flattens and may slightly reverse at very high ages of first marriage (late thirties and beyond), where individual selectivity effects and smaller pools of potential partners may introduce their own complications. The most favourable age window for first marriage in terms of longevity outcomes appears to be the mid-to-late twenties, when education is typically complete, financial independence is established, and identity is more stable.

Research on adult attachment theory, building on Bowlby's original framework and developed by Hazan and Shaver (1987), finds consistent links between early attachment patterns and adult relationship outcomes. Adults with secure attachment styles (approximately 55-60% of the population in Western samples) report higher relationship satisfaction, better conflict resolution, and longer relationship durations on average. Adults with anxious attachment styles show elevated rates of relationship distress and relationship-seeking behaviour after dissolution. Avoidant attachment is associated with lower commitment and higher rates of initiated dissolution. Importantly, attachment style is not fixed: secure relationships and therapy can shift insecure attachment patterns toward greater security over time.

A systematic review by Stafford (2010, Journal of Communication) found that long-distance relationships do not show lower satisfaction or commitment than geographically proximate relationships, and in some studies show higher idealisation and intimacy during the distance phase. However, the transition from long-distance to cohabiting is a period of elevated dissolution risk: the "transition turbulence" effect, first documented by Sahlstein (2004), describes how couples who have adapted to distance sometimes struggle with the adjustments required by proximity. Studies estimate that approximately one-third of long-distance relationships do not survive within three months of closing the distance. Planned transition timelines and realistic expectations about the adjustment period are associated with better outcomes.

CDC National Vital Statistics data finds that the median duration of marriages ending in divorce in the United States is approximately 8 years. This figure masks substantial variation: marriages that end in the first three years are usually distinct from longer-term dissolutions in their causes and profiles. UK ONS data for 2023 found a higher median of 12.7 years for divorcing couples in England and Wales, reflecting both later marriage ages and different cultural and legal patterns around divorce. It is worth noting that these figures describe only marriages that dissolve, not all marriages: the full population of marriages, including those that persist, has a very different and much longer median duration.