The efficacy of thrombolytic therapy for acute myocardial infarction depends partly on how soon after the onset of symptoms it is administered. We therefore studied the efficacy and safety of thrombolytic therapy administered before hospital admission and thrombolytic therapy administered after admission in patients with suspected myocardial infarction. In a multicenter, double-blind study, patients seen within six hours of the onset of symptoms who had a qualifying 12-lead electrocardiogram were randomly assigned to receive either anistreplase before admission, followed by placebo in the hospital (prehospital group), or placebo before admission, followed by anistreplase in the hospital (hospital group). Prehospital therapy was administered by emergency medical personnel. A total of 2750 patients were randomly assigned to the prehospital group, and 2719 to the hospital group. The patients in the prehospital group received thrombolytic therapy a median of 55 minutes earlier than those in the hospital group. We observed a nonsignificant reduction in overall mortality at 30 days in the prehospital group (9.7 percent vs. 11.1 percent in the hospital group; reduction in risk, 13 percent; 95 percent confidence interval, -1 to 26 percent; P = 0.08). Death from cardiac causes was significantly less frequent in the prehospital group than in the hospital group (8.3 percent vs. 9.8 percent; reduction in risk, 16 percent; 95 percent confidence interval, 0 to 29 percent; P = 0.049). Particular adverse events occurred more frequently in the prehospital group during the period before hospitalization; among these events were ventricular fibrillation (P = 0.02), shock (P < 0.001), symptomatic hypotension (P < 0.001), and symptomatic bradycardia (P = 0.001). With the exception of symptomatic hypotension, however, the overall incidence of these events was similar for both groups. Prehospital thrombolytic therapy for patients with suspected myocardial infarction is both feasible and safe when administered by well-equipped, well-trained mobile emergency medical staff. Although such therapy appears to reduce mortality from cardiac causes, our data do not definitely establish that it reduces overall mortality.