OBJECTIVE: To determine the short-term effects of using genotypic antiretroviral resistance testing (GART) with expert advice in the management of patients failing on a protease inhibitor and two nucleoside reverse transcriptase inhibitors. DESIGN: Prospective randomized controlled trial. SETTING: Multicenter community-based clinical trials network. PATIENTS: One-hundred and fifty-three HIV-infected adults with a threefold or greater rise in plasma HIV-1 RNA on at least 16 weeks of combination antiretroviral therapy. INTERVENTIONS: Randomization was either to a GART group, where genotype interpretation and suggested regimens were provided to clinicians, or to a no-GART group, where treatment choices were made without such input. MAIN OUTCOMES MEASURES: Plasma HIV-1 RNA levels and CD4 cell counts were measured at 4, 8, and 12 weeks following randomization. The primary endpoint was change in HIV-1 RNA levels from baseline to the average of the 4 and 8 week levels. RESULTS: The average baseline CD4 cell count was 230 x 10(6) cells/l and the median HIV-1 RNA was 28,085 copies/ml. At entry, 82 patients were failing on regimens containing indinavir, 51 on nelfinavir, 11 on ritonavir, and nine on saquinavir. HIV-1 RNA, averaged at 4 and 8 weeks, decreased by 1.19 log10 for the 78 GART patients and -0.61 log10 for the 75 no-GART patients (treatment difference: -0.53 log, 95% confidence interval, -0.77 to -0.29; P = 0.00001). Overall, the best virologic responses occurred in patients who received three or more drugs to which their HIV-1 appeared to be susceptible. CONCLUSION: In patients failing triple drug therapy, GART with expert advice was superior to no-GART as measured by short-term viral load responses.