Accurate monitoring of the effectiveness of protected areas (PAs) in decreasing deforestation is increasingly important given the vital role of forest protection in climate change mitigation. Recent studies on PA effectiveness have used remote-sensing imagery to compare deforestation rates within PAs to surrounding areas. However, remote-sensing data used in isolation provides limited information on the factors contributing to effectiveness. We used landscape-modelling techniques to estimate the effectiveness of ten PAs in Madre de Dios, Peru. Factors influencing PA effectiveness were investigated using in situ key-informant interviews. Although all of the PAs studied had positive effectiveness scores, those with the highest scores were ecotourism and conservation concessions, where monitoring and surveillance activities and good relations with surrounding communities were reported as possible factors in decreasing deforestation rates. Native community areas had the lowest scores, with deforestation mainly driven by internal resource use and population growth. Weak local governance and immigration were identified as underlying factors decreasing the effectiveness of protection, whereas good relations with surrounding communities and monitoring activity increased effectiveness. The results highlight the need to combine remote sensing with in situ information on PA management because identification of drivers and deterrents of deforestation is vital for improving the effectiveness of protection.