OBJECTIVE To compare the safety and activity of lamivudine plus zidovudine with the safety and activity of zalcitabine plus zidovudine in patients with moderately advanced human immunodeficiency virus (HIV) infection who had received zidovudine. METHODS A multicenter, randomized, double-blind, three-arm, 24-week study with a blinded extension through at least 52 weeks. METHODS 21 sites in the United States, Canada, and Puerto Rico. METHODS 254 patients who had received zidovudine (median duration of previous therapy, 20 months) and had absolute CD4+ cell counts of 100 to 300 cells/mm3. METHODS Patients were randomly assigned to receive one of three regimens: 150 mg of lamivudine twice daily plus 200 mg of zidovudine three times daily (low-dose lamivudine group); 300 mg of lamivudine twice daily plus 200 mg of zidovudine three times daily (high-dose lamivudine group); or 0.75 mg of zalcitabine plus 200 mg of zidovudine three times daily (zalcitabine group). METHODS Immunologic activity was assessed primarily by changes in absolute CD4+ cell counts; virologic activity was assessed by changes in plasma HIV RNA levels as measured by reverse transcriptase polymerase chain reaction. Safety of the treatment regimens was assessed through the reporting of adverse events. RESULTS 78% of patients completed 24 weeks of study treatment, and 63% of patients completed 52 weeks of study treatment. Changes in absolute CD4+ cell counts were significantly better for the low-dose and the high-dose lamivudine groups than for the zalcitabine group (median changes at 52 weeks were +42.5 cells/mm3 in the low-dose lamivudine group, +23.33 cells/mm3 in the high-dose lamivudine group, and -29.58 cells/mm3 in the zalcitabine group). Suppression of plasma HIV RNA levels was similar for all groups (median changes at 52 weeks were -0.48 log10 copies/mL in the low-dose lamivudine group, -0.51 log10 copies/mL in the high-dose lamivudine group, and -0.39 log10 copies/mL in the zalcitabine group). No significant differences in safety were seen among the three regimens, although the low-dose lamivudine regimen appeared to be better tolerated than the others. CONCLUSIONS In patients with HIV infection who had previously received zidovudine, 150 mg of lamivudine plus zidovudine resulted in greater immunologic evidence of benefit than did 0.75 mg of zalcitabine plus zidovudine and was better tolerated than 300 mg of lamivudine plus zidovudine.