Computer linkage of obstetric and psychiatric record systems made it possible to identify all women resident in the city of Edinburgh who had given birth to live or stillborn children in 1971-7 and to study (a) the distribution of psychiatric admissions relative to the time of childbirth and (b) the correlates of psychiatric admission in the first 90 days after childbirth. Having a first baby, being unmarried and undergoing Cesarean section were all associated with an increased risk of admission; twin births, perinatal death and maternal age were not. It is difficult to account for these and other established relationships purely in psychological or purely in somatic terms, suggesting that influences of both kinds are probably involved in the genesis of puerperal disorders.