The low-latitude regions of the Earth are undergoing profound, rapid landscape change as forests are converted to agriculture to support growing population. Understanding the effects of these land-use changes requires analysis of watershed-scale geomorphic processes to better inform and manage this usually disorganized process. The investigation of hillslope erosion and the development of sediment budgets provides essential information for resource managers. Four small, montane, humid-tropical watersheds in the Luquillo Experimental Forest and nearby Río Grande de Loíza watershed, Puerto Rico (18° 20' N., 65° 45' W.), were selected to compare and contrast the geomorphic effects of land use and bedrock geology. Two of the watersheds are underlain largely by resistant Cretaceous volcaniclastic rocks but differ in land use and mean annual runoff: the Mameyes watershed, with predominantly primary forest cover and runoff of 2,750 millimeters per year, and the Canóvanas watershed, with mixed secondary forest and pasture and runoff of 970 millimeters per year. The additional two watersheds are underlain by relatively erodible granitic bedrock: the forested Icacos watershed, with runoff of 3,760 millimeters per year and the agriculturally developed Cayaguás watershed, with a mean annual runoff of 1,620 millimeters per year. Annual sediment budgets were estimated for each watershed using landslide, slopewash, soil creep, treethrow, suspended sediment, and streamflow data. The budgets also included estimates of sediment storage in channel beds, bars, floodplains, and in colluvial deposits. In the two watersheds underlain by volcaniclastic rocks, the forested Mameyes and the developed Canóvanas watersheds, landslide frequency (0.21 and 0.04 landslides per square kilometer per year, respectively), slopewash (5 and 30 metric tons per square kilometer per year), and suspended sediment yield (325 and 424 metric tons per square kilometer per year), were lower than in the two watersheds underlain by granitic bedrock. In these granitic watersheds, landslide frequency, slopewash, and suspended sediment yield were 0.43 landslides per square kilometer per year, 20 metric tons per square kilometer per year, and 2,140 metric tons per square kilometer per year, respectively, in the forested Icacos watershed and 0.8 landslides per square kilometer per year, 105 metric tons per square kilometer per year, and 2,110 metric tons per square kilometer per year, respectively, in the agriculturally developed Cayaguás watershed. Comparison of sediment budgets from the forested and developed watersheds indicates that human activities increase landslide frequency by as much as factor of 5 and slopewash by as much as a factor of 6. When the difference in annual runoff is considered, the effect of land use on suspended sediment yields is also notable. Sediment concentration, calculated as sediment yield normalized by runoff, was about 2.3 to 3.7 times as great in the two watersheds in secondary forest and pasture compared with sediment concentration in the watersheds in primary forest. Even in the two watersheds with primary forest cover, the Mameyes and Icacos, located in the Luquillo Experimental Forest, the effects of anthropogenic disturbance were marked: 43 to 63 percent of landslide-related erosion was associated with road construction and maintenance.