Résumé : The sensitivity and discriminatory power of the 151 and 215 amplification refractory mutation system (ARMS) were evaluated, and their performance for the detection of drug resistance in mixed genotypic populations of the reverse transcription (RT) gene of HIV-1 were compared with T7 sequencing, cycle sequencing, the line probe assay (LiPA) HIV-1 RT test, and the recombinant virus assay (RVA). ARMS and the LiPA HIV-1 RT test were shown to be able to detect minor variants that in particular cases comprised only 1%. T7 sequencing on an ALF semiautomated sequencer could correctly score mixtures only when variants were present at 50%. Cycle sequencing on an ABI PRISM 310 improved the sensitivity for mixtures to about 25%. Using RVA, it was shown that at least 50% of the virus population needed to carry the resistance mutation at codon 184 to afford phenotypic resistance against lamivudine. The two point mutation assays therefore proved to be more sensitive methods than sequencing and RVA to reliably determine a gradual shift in HIV-1 drug resistance mutations in follow-up of patients infected with HIV-1. In 4 of 5 treated patients who were followed by ARMS, a gradual shift in resistant genotypic populations was observed during a period of 6 to 19 months. For 1 patient, a shift from wild to mutant type at position 151 occurred within 2 months, without mixed genotypic intermediate type's being detected.