We randomly assigned 208 patients who underwent brain biopsy for presumptive herpes simplex encephalitis to receive either vidarabine (15 mg per kilogram of body weight per day) or acyclovir (30 mg per kilogram per day) for 10 days. Sixty-nine patients (33 percent) had biopsy-proved disease; 37 received vidarabine, and 32 acyclovir. The mortality in the vidarabine recipients was 54 percent, as compared with 28 percent in the acyclovir recipients (P = 0.008). Six-month mortality varied according to the Glasgow coma score at the onset of therapy. For scores of greater than 10, 7 to 10, and less than or equal to 6, mortality was 42, 46, and 67 percent in the patients treated with vidarabine, as compared with 0, 25, and 25 percent in those treated with acyclovir. A six-month morbidity assessment using an adapted scoring system revealed that 5 of 37 patients receiving vidarabine (14 percent) as compared with 12 of 32 receiving acyclovir (38 percent) were functioning normally (P = 0.021). Eight vidarabine-treated patients (22 percent) and three acyclovir-treated patients (9 percent) had moderate debility. Patients under 30 years of age and with a Glasgow coma score above 10 had the best outcome with acyclovir treatment. We conclude that acyclovir is currently the treatment of choice for biopsy-proved herpes simplex encephalitis.