An in-house genotypic antiretroviral resistance assay was evaluated by testing 32 plasma samples obtained from heavily pretreated human immunodeficiency virus type 1 (HIV-1)-infected patients failing multiple antiretroviral regimens. The same samples were also sent to Virco Laboratories for genotypic (VircoGEN) and phenotypic (Antivirogram) resistance analysis. Sequencing results obtained by in-house (HG) and VircoGEN (VG) genotyping were concordant for 387 of 400 (96.75%) drug resistance mutations. Genotype-based prediction of drug susceptibility for 13 currently licensed antiretroviral compounds were in agreement in 336 (80.78%) cases, partially concordant in 73 (17.54%) cases and discordant in only seven (1.68%) cases. VG indicated 'possible resistance' twice as much as HG. When genotype interpretation was compared with the Antivirogram phenotypic data, there were 27 (6.49%) and 23 (5.52%) wrong calls by HG and by VG, respectively. Both assays were more sensitive in detecting drug resistance than drug susceptibility (94.61 vs. 65.19% for HG, 80.84 vs. 56.91% for VG) and more specific in detecting drug susceptibility than drug resistance (93.62 vs. 73.49% for HG, 93.62 vs. 80.32% for VG). Rule-based algorithms can reliably interpret genotypic data obtained from most heavily pretreated patients. However, occasional genotypic patterns may be erroneously interpreted without resistance phenotyping.