Zidovudine is approved for administration in doses given every 4 hours. Less frequent dosing has been used in many clinical trials, but the toxicity and efficacy of such regimens have not been formally compared with the approved regimen. In this multicenter, randomized, double-blind, controlled trial, the safety, tolerance and efficacy of 600 mg of zidovudine given daily in two or six divided doses were compared. Three hundred and twenty patients with a CD4 lymphocyte count < 250 cells/mm3 (mean, 104 cells/mm3) or a prior AIDS-defining illness were treated with zidovudine 100 mg every 4 hours (regimen A) or 300 mg every 12 hours (regimen B). Eighty-eight patients (56%) and 94 patients (58%), assigned to regimens A and B, respectively, completed the planned 48 weeks of treatment. Serious anemia (hemoglobin < or = 7.5 g/dl) occurred in 13% and 7% of patients treated with regimens A and B, respectively (difference, 6%, 95% confidence interval [CI], 2, 12%; p = .13). The mean duration of treatment and the frequency of neutropenia and symptomatic complaints including nausea and headache were similar in the two treatment groups. The number of patients experiencing a new opportunistic infection (18% versus 20% for regimens A and B, respectively), and the number of deaths (five in each group) did not differ significantly between groups. The effect of treatment on CD4 lymphocyte counts and HIV p24 antigenemia also was similar for both regimens. Zidovudine given at the more convenient dose of 300 mg twice daily has similar safety, and tolerance and appears to have similar efficacy to the currently approved regimen. Use of this regimen should help simplify the treatment of HIV disease.