Computer linkage of an obstetric register and a psychiatric case register made it possible to investigate the temporal relationship between childbirth and psychiatric contact in a population of 470 000 people over a 12-year period: 54 087 births resulted in 120 psychiatric admissions within 90 days of parturition [corrected]. The 'relative risk' of admission to a psychiatric hospital with a psychotic illness was extremely high in the first 30 days after childbirth, particularly in primiparae, suggesting that metabolic factors are involved in the genesis of puerperal psychoses. However, being unmarried, having a first baby, Caesarian section and perinatal death were all associated with an increased risk of psychiatric admission or contact, or both, suggesting that psychological stresses also contribute to this high psychiatric morbidity. Women with a history of manic depressive illness, manic or depressive, had a much higher risk of psychiatric admission in the puerperium than those with a history of schizophrenia or depressive neuroses, and the majority of puerperal admissions met Research Diagnostic Criteria for manic or depressive disorder. Probably, therefore, puerperal psychoses are manic depressive illnesses and unrelated to schizophrenia.