Hospital records of patients with asthma admitted to teaching hospitals in Perth, Western Australia between 1976 and 1980 were examined retrospectively to identify characteristics of the illness which were associated with subsequent death. From 5722 admissions there were 195 deaths to December 1982, 186 of whom had records available (cases); 452 of the surviving subjects were used for comparison (controls). There was no difference in age of onset of asthma or cigarette smoking habits between the two groups, but ischaemic heart disease as an associated condition was significantly more frequent in cases. On admission to hospital an arterial PCO2 less than 45 mmHg was more frequent in those who died, but there were no differences in arterial PO2, lowest pH, highest or lowest FEV1 and FVC. Cases more frequently used home nebulisers and were more frequently prescribed corticosteroids, antibiotics and sedatives or tranquilizers prior to admission, corticosteroids and sedatives or tranquilisers during admission and sedatives or tranquilisers on discharge. These results suggest that cases had more severe asthma in that they were more often treated with home nebulisers, corticosteroids and antibiotics, but with the exception of PaCO2 the commonly used measurements of severity of asthma did not identify those at risk of death. The prescription of sedatives or tranquillisers appears to be associated with an increased risk of death in subjects with asthma.