To evaluate the antiretroviral activity of delavirdine mesylate, a non-nucleoside reverse transcriptase inhibitor of HIV-1, we performed a phase II, randomized, double-blind, multicenter trial comparing the three-drug combination of delavirdine with zidovudine and didanosine to two-drug combinations of these drugs. Patients with CD4 cell counts between 100 and 500 cells/mm3 without prior or <6 months of monotherapy with zidovudine or didanosine were randomized to one of four arms and observed on a follow-up basis for 48 weeks. In total, 544 patients were evaluated. In those assigned to the three-drug regimen, mean short-term (weeks 4-12) and long-term (weeks 40-48) change in CD4 cells from baseline were 49.3+/-8.1 and 65.4+/-13.4 cells/mm3, respectively; mean short-term and long-term HIV-1 RNA changes from baseline were -1.13 log10+/-0.12 and -0.73+/-0.12 copies/ml, respectively. These responses in CD4 cell counts and HIV-1 RNA levels were better in comparisons with each of the two-drug arms at all study points; however, differences were not consistently significant. Gastrointestinal side effects were experienced by 33% of patients (178 of 544), and 30% (121 of 407) receiving delavirdine experienced rash, only one case of which was severe. In this study, therapy with delavirdine + zidovudine + didanosine was safe and showed modest, but not always significant, antiviral activity and CD4 cell count benefit compared with two-drug regimens with these agents. Key