OBJECTIVE The aims of this study were to: (i) determine the incidence of thymidine-associated mutations (TAMs) in an observational clinical cohort of naive HIV-1 patients who stopped first-line therapy including either zidovudine or stavudine; and (ii) assess the immunological and virological responses to subsequent second-line therapy in patients who switched from zidovudine to stavudine or conversely. METHODS Plasma samples from 165 patients who stopped first-line antiretroviral therapy containing either zidovudine or stavudine were examined for the presence of drug-resistant genotypes. Subsequent second-line immunological and virological follow-up was performed in 136 patients who switched from zidovudine to stavudine and conversely. RESULTS Among the 93 patients who stopped first-line therapy including zidovudine and the 72 who stopped first-line therapy including stavudine, genotypic resistance testing was available for 67 (72%) and 54 (75%), respectively. The presence of TAMs was significantly more frequent in the zidovudine than the stavudine group (23.8% versus 5.5; P = 0.006). The short- and long-term immunological and virological responses to second-line therapy were comparable in the zidovudine and stavudine groups, despite different baseline profiles of viral resistance (median increase in CD4 cells/mm3 at 1 year of therapy, 118 versus 119; viral load <400 copies/mL, 47% versus 47%). CONCLUSIONS These results suggest that TAMs occur in more patients on antiretroviral regimens including zidovudine than on regimens including stavudine. Although the results from observational studies should be interpreted cautiously, these findings may be useful in determining the optimal sequencing of zidovudine and stavudine for the treatment of naive HIV-1-infected patients.