A prospective, stratified, randomized clinical trial of the safety and efficacy of four general anesthetic agents (enflurane, fentanyl, halothane, and isoflurane) was conducted in 17,201 patients (study population). Patients were studied before, during, and after anesthesia for up to 7 days. Nineteen patients died (0.11%), and in seven of these (0.04%) the anesthetic may have been a contributing factor. The rates of death, myocardial infarction, and stroke in the study population were so low (less than 0.15%) that no conclusions regarding the relative rates of these outcomes among the four anesthetic agents could be reached. The rates of 16 of 66 types of adverse outcomes in the study population were significantly different among the four study agents. Most of these outcomes were minor. However, severe ventricular arrhythmia (P less than 10(-6)) was more common with halothane, severe hypertension (P less than 10(-6)) and severe bronchospasm (P = 0.028) were more common with fentanyl, and severe tachycardia (P = 0.001) was more common with isoflurane. Recovery from anesthesia during the first 30 min was slowest in those patients who received halothane (P less than or equal to 0.001). In addition, patients who received fentanyl experienced less pain during the first hour in the recovery room (P less than 10(-6)). In conclusion, clinically important differences do exist for some outcomes among the four study agents.