Both acyclovir and vidarabine are effective treatment for varicella zoster virus (VZV) infection in immunosuppressed patients. To determine which is preferable, therapy with these two agents was compared in a prospective, randomized trial. A total of 22 immunocompromised patients undergoing treatment for hematologic malignancies and presenting with VZV infection within 72 hours of the onset of rash were randomly assigned to receive intravenous acyclovir or vidarabine; 11 patients were randomly assigned to each treatment group. Acyclovir was significantly more effective than vidarabine in preventing complications of VZV infection, and treatment failures requiring a change to the alternate therapy occurred only among those treated with vidarabine. As compared with vidarabine, acyclovir shortened the median period during which results of viral culture specimens were positive and new lesions formed. Acyclovir also shortened the median interval until the first decrease in pain, the crusting of all lesions, and the complete healing of lesions. Acyclovir is more effective than vidarabine in the treatment of VZV infection in severely immunocompromised patients and should be considered the treatment of choice in such cases.