The resistance profile of darunavir (TMC114) in treatment-experienced patients was explored using pooled week 24 data from POWER 1, 2, and 3 at the recommended dose of darunavir with low-dose ritonavir (darunavir/r, 600/100 mg bid, N = 458). Baseline darunavir fold change in EC(50) was a strong predictor of virological response at week 24. Preliminary phenotypic clinical cut-offs of 10 and 40 were established. Virological response to darunavir/r was maintained in the presence at baseline of a high number of IAS-USA PI resistance-associated mutations (IAS-USA PI RAMS); a diminished response occurred with >or=14. Eleven protease mutations associated with diminished darunavir/r virological response were identified (V11I, V32I, L33F, I47V, I50V, I54L/M, G73S, L76V, I84V, and L89V). These darunavir resistance-associated mutations (DRV RAMS) occurred in the presence of a high number of IAS-USA PI RAMS. Virological response was diminished with three or more DRV RAMS in the background of a high number of IAS-USA PI RAMS. Incremental numbers of DRV RAMS were more predictive of outcome than were IAS-USA PI RAMS. Mutations developing during darunavir/r virological failure (V32I, L33F, I47V, I54L, and L89V) were also featured in the DRV RAMS list. Site-directed mutants carrying these five mutations, or any one of these mutations either alone or together with one or two IAS-USA PI RAMS, showed no reduced darunavir susceptibility, suggesting that a high number of additional background mutations is required for darunavir resistance. In this population of treatment-experienced patients, darunavir/r demonstrated significantly greater efficacy than investigator-selected control PIs of trials POWER 1 and 2, regardless of baseline viral genotype or phenotype, while exhibiting a high genetic barrier to the development of resistance.