In the ALBI trial, 151 antiretroviral-naive patients with plasma human immunodeficiency virus type 1 (HIV-1) RNA levels of 10,000 to 100,000 copies/ml and CD4 cell counts > or = 200 cells/mm3 received 24 weeks of treatment with stavudine/didanosine (n=51), zidovudine/lamivudine (n=51) or stavudine/didanosine for 12 weeks followed by zidovudine/lamivudine (n=49). Baseline plasma HIV-1 RNA and CD4 cell counts were comparable in the treatment groups. The mean decrease in plasma HIV-1 RNA at 24 weeks in the stavudine/didanosine group (2.26 log10) was significantly greater than that in either the zidovudine/lamivudine group (1.26 log10) or the alternating treatment group (1.58 log10) (P<0.0001 for both). Proportions of patients with plasma HIV-1 RNA level <500 copies/ml (91% vs 42% and 60%) and <50 copies/ml (47% versus 4% and 9%) were significantly greater in the stavudine/didanosine group (P<0.001 for pairwise comparisons). Stavudine/didanosine was associated with a mean increase in CD4 cell count (124 cells/mm3) significantly greater than that in the zidovudine/lamivudine group (62 cells/mm3, P<0.01) and comparable to that in the alternating group (118 cells/mm3). All study regimens were well tolerated. These findings, indicating superiority of stavudine/didanosine over zidovudine/lamivudine in virological and immunological response over 24 weeks, suggest that the combination should be considered as a basis for highly active antiretroviral therapy.