We assessed the performance of HIV-1 genotyping tests in rescue therapy. Patients were divided into two groups: group 1 (genotyped), included those switching to new antiretroviral drugs based on HIV-1 genotyping data, and group 2 (standard of care -SOC), comprised those in rescue therapy who had not used this test. This was an open and non-randomized study, with 74 patients, followed up for a mean period of 12 months, from February 2002 to May 2003. The groups differed in the duration of antiretroviral use, experience with diverse drug classes (non-nucleoside reverse transcriptase inhibitors and protease inhibitors) and viral load <2.6 log 10 copies/mL at any time during treatment. In 23 patients (group 1), the switch in antiretroviral (ARV) regimen was based on genotyping data; this test was not used for 51 patients (group 2). Two CD 4 + lymphocyte counts and viral load counts were made for each patient during the study. Data from the pharmacy where patients received antiretroviral agents, medical charts, and direct interviews with patients to assess compliance to treatment, were analyzed. In the genotyped group, the average drop in viral load was 2.8 log 10 , compared with a 1.5 log 10 difference in group 2; the difference was significant in the first assessment performed six months after switching (p=0.001). Considering the patients with viral load < 2.6 log 10 (400 copies/mL) after switching, the patients in group 1 had a better performance in the first assessment (73.9% versus 31.1% in groups 1 and 2, respectively); this difference was significant (p=0.001). In multivariate analysis, the variables associated with a greater drop in viral load in the first assessment were the patients whose switching was based on genotyping (group 1), those with a past history of viral load < 2.6 log 10 and correct use of antiretroviral agents. In conclusion, the genotyping test and adherence were found to be independent factors for success in the management of patients who failed treatment.