BACKGROUND: Growing evidence has linked HIV-1 resistance mutations and drug failure. The use of genotypic-resistance analysis to assist therapeutic decision-making in patients failing therapy has not been investigated. We assessed the virological and immunological impact of genotypic-resistance testing. METHODS: We did a prospective, open, randomised, controlled study of HIV-1-infected patients in whom combination therapy was not successful. We randomly assigned patients standard care (control, n=43) or treatment according to the resistance mutations in protease and reverse-transcriptase genes (genotypic group, n=65). The major endpoint was the change in HIV-1 RNA viral load. Analysis was by intention to treat. FINDINGS: 108 patients were enrolled. All patients were similar for risk factors, age, sex, previous treatment, CD4-cell count (214/microL [SD14]) and log HIV-1 RNA viral load at baseline (4.7 copies/mL [0.1]). At month 3, the mean change in HIV-1 RNA was -1.04 log (0.14) in the study group compared with -0.46 log (0.17) in the control group (mean difference 0.58 log [95% CI 0.14-1.02], p=0.01). At month 6, changes were -1.15 (0.15) log copies/mL, and -0.67 (0.19) log copies/mL in the genotypic group and the control group, respectively (mean difference 0.48 log [0.01-0.97], p=0.05). Difference in the drop in viral load combined at 3 months and 6 months was significant (p=0.015). At month 3, HIV-1 RNA was lower than detection level (200 copies/mL) in 29% (19/65) of patients in the genotypic group versus 14% (6/43) in the control group (p=0.017). At month 6, the values were 32% (21/65) and 14% (6/43) (p=0.067) for the genotypic group and the control group, respectively. Therapy was generally well tolerated, with ten patients (six in the genotypic group, four in the control group) requiring toxic-effect-related drug modification. INTERPRETATION: We found genotypic-resistance testing to have a significant benefit on the virological response when choosing a therapeutic alternative. Further study of the use of genotypic-resistance testing in assisting clinical decision-making is warranted.