To understand more fully the nature of events leading to asthmatic death, we conducted a confidential enquiry prospectively throughout 1994-96 among the surviving relatives and respective general practitioners of subjects whose deaths could be attributed to asthma, whether wholly or partly. We also reviewed relevant hospital records and autopsy reports, and we submitted all the gathered information to an enquiry panel for evaluation. The subjects were identified from death certificates issued in five districts of the Northern Health Region of England (population 1 million) on which asthma was recorded as the primary cause of death. The enquiry panel agreed that asthma had been a critical factor in causing death in only 33 of the 79 certified cases for which there were sufficient data. The level of concordance was substantially greater for subjects aged < 65 years (76%) than for those who were older (17%). In 16 of the 33 cases asthma alone appeared to be responsible for death, but in 17 cases a wide variety of additional, co-morbid, disorders appeared to have contributed. They included, during the 24 h preceding death, gastric aspiration, septicaemia, a single dose of a beta-blocker, the abuse of organic solvents or illicit drugs and possibly, an inadvertent exposure to horse allergen. More chronic causes of co-morbidity included ischaemic heart disease, chronic obstructive pulmonary disease (COPD), thoracic cage deformity and alcohol abuse. There were possible errors of judgement in two cases by the supervising physician (6%) and in three cases by the patient (9%). Poor compliance and psychosocial disruption probably exerted an additional adverse influence in nine cases (27%). We conclude: (1) that asthma death certification in subjects aged 65 years or more is very unreliable, (2) that for approximately half of the deaths in which asthma exerted a critical role there were critical co-morbid disorders and (3) that errors of judgement, poor compliance, or psychosocial disruption are likely to have exerted an additional adverse influence in an important minority of cases.