Despite the recent development of apparently effective asthma drugs, the number of deaths from asthma has not declined. The authors tried to establish an optimal strategy for the prevention of acute asthmatic death by analyzing the circumstances of acute fatal or near-fatal asthma. Data were collected from 51 adult patients admitted to Bokutoh Tertiary Emergency Center due to acute asthma between November 1985 and May 1990 and 38 asthmatic patients admitted to Yokohama City Hospital in 1990. Pre-admission data were obtained through interviews with the patients, their families, or doctors who had seen them. A total of 89 patients were classified into three subgroups: group 1 consisted of patients dead-on-arrival (DOA); group 2, non-DOA patients with disturbed consciousness; and group 3, patients with less severe episodes. Little background information was significantly different among groups, but symptomatic episodes in group 1 patients tended to occur more rapidly. The speed of onset of the episodes was also dependent on the asthma control status. Prehospital care of groups 1 and 2 patients was very poor despite severity of the symptoms. All patients in groups 2 and 3 were successfully treated and discharged, but five of the 26 patients in group 2 died during follow-up periods. Prognosis of patients after discharge appeared to be dependent upon asthma control status before the acute episodes. It was concluded that acute fatal or near fatal asthma could occur in apparently low-risk patients as well as high-risk ones. It was also suggested that the optimal strategy for preventing asthma deaths might be variable.