Thirty-four subjects with symptomatic HIV-1 infection, p24 antigenaemia, and CD4 cell counts > 200/mm3 were randomly assigned to receive treatment with either zidovudine (ZDV) orally, interferon-alpha (IFN-alpha) subcutaneously, or both at respective low (200 mg ZDV/ 2 million international units IFN-alpha (MIU)), middle (400 mg/4 MIU) or high (600 mg/6 MIU) daily dose levels for 12 weeks. Thereafter, all patients received combination therapy at the initially assigned dose level to a total of 96 weeks. This design permitted analysis by the combination index (CI) method, which demonstrated antiretroviral synergy between ZDV and IFN-alpha with respect to p24 antigen suppression. Over the first 12 weeks, combination therapy was acceptably tolerated, more so than IFN-alpha monotherapy, and it was significantly more active in suppressing antigenaemia than either of the monotherapies. Similarly, the high-dose combination was the most active dose level over weeks 12 to 96. Combination ZDV/IFN-alpha at the optimal dose level defined by this trial merits further study. In addition, the CI design strategy employed here may be useful for the investigation of new antiretroviral combinations.