We assessed mortality rates over 25 years in 212 patients admitted for depression or depressive symptoms. More patients had died than expected (80 including 13 suicides; SMR = 1.40, P < 0.01). Females, but not males, experienced significantly higher mortality than the general population. When suicides were excluded, neither the group as a whole, nor females solely, demonstrated excess mortality. Rigorously diagnosed depressive sub-type did not predict mortality. Mortality, particularly from suicide, was disproportionately greater in the first 2 years after index admission, suggesting that the follow-up of patients hospitalised with depression must be especially assiduous during the years immediately after admission.