OBJECTIVES: To describe the use of second-line protease-inhibitor regimens in Médecins Sans Frontières HIV programmes, and determine switch rates, clinical outcomes, and factors associated with survival. DESIGN/METHODS: We used patient data from 62 Médecins Sans Frontières programmes and included all antiretroviral therapy-naive adults (> 15 years) at the start of antiretroviral therapy and switched to a protease inhibitor-containing regimen with at least one nucleoside reverse transcriptase inhibitor change after more than 6 months of nonnucleoside reverse transcriptase inhibitor first-line use. Cumulative switch rates and survival curves were estimated using Kaplan-Meier methods, and mortality predictors were investigated using Poisson regression. RESULTS: Of 48,338 adults followed on antiretroviral therapy, 370 switched to a second-line regimen after a median of 20 months (switch rate 4.8/1000 person-years). Median CD4 cell count at switch was 99 cells/microl (interquartile ratio 39-200; n = 244). A lopinavir/ritonavir-based regimen was given to 51% of patients and nelfinavir-based regimen to 43%; 29% changed one nucleoside reverse transcriptase inhibitor and 71% changed two nucleoside reverse transcriptase inhibitors. Median follow-up on second-line antiretroviral therapy was 8 months, and probability of remaining in care at 12 months was 0.86. Median CD4 gains were 90 at 6 months and 135 at 12 months. Death rates were higher in patients in World Health Organization stage 4 at antiretroviral therapy initiation and in those with CD4 nadir count less than 50 cells/microl. CONCLUSION: The rate of switch to second-line treatment in antiretroviral therapy-naive adults on non-nucleoside reverse transcriptase inhibitor-based first-line antiretroviral therapy was relatively low, with good early outcomes observed in protease inhibitor-based second-line regimens. Severe immunosuppression was associated with increased mortality on second-line treatment.