Forty-five subjects with symptomatic human immunodeficiency virus type 1 (HIV-1) infection, CD4+ lymphocyte counts of > or = 150 x 10(6)/L, and Karnofsky scores > or = 60 were enrolled in a multicenter, randomized, controlled trial that compared zidovudine monotherapy and combination therapy for 48 weeks with zidovudine and interferon-alpha (IFN-alpha). Zidovudine with IFN-alpha (n = 25) had a favorable effect on CD4+ cell counts compared with zidovudine alone (n = 20). At all time points analyzed, the mean change from baseline was higher, reaching significance at week 24 (+10% versus -21%; P = .029). At week 48 the difference was -12% versus -45% (P = .07). Anti-CD3 monoclonal antibody-induced T cell reactivity improved temporarily in both groups. Serum HIV p24 antigen levels decreased maximally during the first 12 weeks of treatment. At weeks 0 and 48, polymerase chain reaction analysis for mutations at codons 67 and 215 of the HIV-1 reverse transcriptase gene conferring zidovudine resistance was conducted in 10 subjects receiving zidovudine and in 8 subjects receiving combination therapy. At week 48, 1 of 8 and 4 of 6 samples from the groups receiving zidovudine only or combination therapy, respectively, contained wild type virus at codon 215. Grade 3 or 4 toxicity was uncommon. Drug-related malaise and anorexia were observed more frequently in patients receiving both zidovudine and IFN-alpha.