Background. Evidence is conflicting as to whether the association between marital status and psychological distress is due to selection (i.e. distress influences marital status) or causation (marital status influences distress). We investigate: (i) whether differences in psychological distress pre-date marital transitions; (ii) whether levels of distress change following transitions; and (iii) potential mediating and moderating factors.

Methods. Data on psychological distress (indicated by the Malaise Inventory) and marital status at ages 23 and 33 were analysed for 4514 men and 4842 women from the 1958 birth cohort.

Results. Higher levels of distress were found among the divorced and lower levels among the single and the married. Selection was seen in the lower initial mean symptoms of those who married (1·69 for men; 2·84 for women) compared to those remaining single (2·41 for men; 3·26 for women). Causation was indicated by the relative deterioration in distress of those who divorced compared to the continuously married (an increase of 0·31 and 0·03 respectively for men), especially in women (a decrease of 0·18 versus 0·71). This was most evident in women who were downwardly mobile and those with children. Recently separated men and women showed especially large increases in distress.

Conclusions. The relationship between marital status and psychological distress involves selection and causation. Findings failed to support ideas of marriage being protective (through social support), or detrimental (through family roles). Divorce increased distress, with both acute and longer-term components moderated by secondary factors such as childcare and declining socio-economic status.