This study evaluated the influence of zidovudine (ZDV) resistance mutations on the antiviral effect of the combination of stavudine (D4T) plus didanosine (ddI) in patients treated previously with ZDV plus zalcitabine (ddC). Twenty patients who had been treated with ZDV plus ddC for a median duration of 11 months (range, 7-42 months) were switched to D4T (40 mg twice a day [BID]) + ddI (200 mg BID) in an open pilot study lasting 6 months. The CDC classes were A (n = 10) and B (n = 10). The median baseline CD4 count was 285/mm(3) and the median baseline plasma virus RNA (Amplicor HIV Monitor RT-PCR assay) was 4.6 log copies/ml. Population-based sequence analysis detected mutations associated with resistance to reverse transcriptase (RT) inhibitors in the RT domain of virus RNA from baseline plasma samples in 13/20 (65%) patients. Twelve patients had mutations associated with zidovudine resistance (3 T215Y - M41L - L210W; 3 T215Y - M41L; 2 T215Y - L210W; 3 T215Y; 1 K70R) and 1 patient had a multi-dideoxynucleoside resistance mutation (QI5IM). Patients with a resistance mutation had a significantly lower RNA suppression after 3 and 6 months (median RNA reduction -0.5 log and -0.1 log) than the remaining patients (-1.6 log and -2 log). Fifty percent of patients with wild-type viruses had undetectable plasma RNA after 24 weeks of D4T plus ddI therapy, whereas all those with mutated viruses had HIV RNA concentration > 3 log copies/ml at week 24 (P <.05). Our finding may have implications when deciding on a second line therapy with three or four drugs that includes two new nucleoside analogues. Cross-resistance between nucleoside analogues deserves maximal attention to ensure optimal antiretroviral therapy and design algorithms for antiretroviral management based on genotypic antiretroviral resistance testing.