Clinical benefit of zidovudine alone in the treatment of HIV infection wanes after several years, with decreasing CD4+ cell numbers and increasing HIV RNA in plasma. To develop treatment strategies following prolonged zidovudine treatment, 92 subjects from the AIDS Clinical Trials Group (ACTG) 175 study after a median of 3.6 years of zidovudine monotherapy were randomized to treatment with stavudine or zidovudine and lamivudine. Evaluation of long-term changes, the average of 40- and 48-week HIV plasma RNA, demonstrated that lamivudine and zidovudine provided significantly greater virologic suppression compared with stavudine (mean decrease 0.70 versus 0.18 1og10 copies/ml,p = 0.003). Twenty-nine percent of zidovudine plus lamivudine recipients had HIV RNA levels below 500 copies per milliliter at 48 weeks as compared with 4% of stavudine recipients (p = 0.02). Both regimens significantly increased CD4+ cell numbers, the means of weeks 40 and 48 rose to 49 and 36 CD4+ cells per cubic millimeter among zidovudine plus lamivudine and stavudine recipients, respectively. Treatments were well tolerated and only 3 of 92 subjects died or developed AIDS within 48 weeks. In zidovudine-experienced subjects, addition of lamivudine resulted in significantly decreased plasma HIV RNA levels at 48 weeks compared with treatment with stavudine alone.