OBJECTIVE: To assess the rate of protocol-defined treatment failure and safety of lopinavir/ritonavir (LPV/r) and saquinavir/ritonavir (SAQ/r). DESIGN: Open-label, prospective, randomized (1:1), international multi-centre trial. METHODS: Adult HIV-1-infected patients were assigned LPV/r 400/100 mg twice daily or SAQ/r 1000/100 mg twice daily with two or more nucleoside reverse transcriptase inhibitors (NRTIs)/non-NRTIs. All patients, whether on or off the assigned treatment, were followed for 48 weeks. RESULTS: Of 339 randomized patients, 324 initiated assigned treatment (intention-to-treat/exposed [ITT/e] population). At 48 weeks, treatment failure occurred in 29/163 (18%) and 53/161 (33%) of patients in the LPV/r and SAQ/r arms, respectively (ITT/e, P = 0.002, log rank test). In an analysis that also considered those patients who discontinued treatment as having failed treatment (ITT/e/discontinuation = failure), 40/161 (25%) LPV/r-treated individuals versus 63/161 (39%) SAQ/R-treated individuals failed treatment (P = 0.005, log rank test). Discontinuation of the assigned treatment occurred in 23/163 (14%) patients in the LPV/r-treated group, compared with 48/161 (300%) in the SAQ/r-treated group (ITT/e; P = 0.001). The primary reasons for premature discontinuation were non-fatal adverse events (LPV/r: 12/163; SAQ/r: 21/161) and patients' choice (LPV/r: 7/163; SAQ/r: 8/161). In the on-treatment analysis of time to treatment failure, no difference was observed between the two arms (P = 0.27, log rank test). CONCLUSION: LPV/r had better antiretroviral effects compared with SAQ/r at the doses and in the formulations studied. This may have been a result of patients' preferences and ability to adhere to assigned therapy, rather than a result of differences in the intrinsic potency of the study protease inhibitors.