There is controversy about the role of beta-agonists in asthma mortality, and the impact of asthma management plans remains unclear. We compared blood beta-agonist levels in patients dying from asthma with those in controls, and estimated the risks associated with specific classes of medication and patterns of management. We identified 89 asthma deaths and recruited 322 patients presenting to hospitals with acute asthma. A questionnaire was administered to the next of kin in 51 cases, and to 202 controls. Blood drawn from 35 cases and 229 controls was assayed for salbutamol. Smoking, drinking, and family problems were significantly more likely among the cases of asthma death than among the controls. The two groups were reasonably well matched with regard to markers of chronic asthma severity. Cases of asthma death were significantly less likely than controls to use a peak flow meter. Written action plans were associated with a 70% reduction in the risk of death. Use of nebulized bronchodilators or oral steroids was significantly more likely in cases of asthma death. Mean blood salbutamol concentrations were 2.5 times higher in cases of asthma. The use of oral steroids for an attack of asthma reduced the risk of death by 90%. More widespread adoption of written asthma management plans, with less reliance on beta-agonists and closer medical supervision, should reduce asthma mortality.