In several clinical trials of interventions designed to lower plasma cholesterol, reductions in coronary heart disease mortality have been offset by an unexplained rise in suicides and other violent deaths. We have tried to find out whether depressive illness is related to low plasma cholesterol concentrations in men of 50 years and older. In 1985-87, Beck depression inventories were obtained from 1020 white men, aged 50-89 years, in the Rancho Bernardo, California, cohort. Disease history and behaviours were assessed by standard questionnaires. Plasma cholesterol and weight were measured at this time, as they had been in 1972-74. Among men aged 70 years and older, categorically defined depression was three times more common in the group with low plasma cholesterol (< 4.14 mmol/L) than in those with higher concentrations (5/31 [16%] vs 22/363 [6%]; p = 0.033). Depressive symptom scores correlated significantly and inversely with plasma cholesterol concentrations, even after adjustment for age, health status, number of chronic illnesses, number of medications, and exercise, as well as measured weight loss and change in plasma cholesterol in the previous 13 years. Our finding that low plasma cholesterol is associated with depressive symptoms in elderly men is compatible with observations that a very low total cholesterol may be related to suicide and violent death. Since cholesterol lowering in the general population is widely recommended, this observation warrants further investigation.