Landslides are common processes in Rio de Janeiro, mainly during heavy summer rainfalls. These natural phenomena normally occur due to a complex combination of natural and human-induced factors. It is evident that the topographic hollows are characterized by landslides hazards because these areas are frequently saturated by water and filled up by colluvial soils. A physically-based model for the topographic control on shallow landsliding, combining a hydrological model and a slope stability model, has been applied to predict landslide hazard areas in an experimental basin. The hydrological model is based on soil properties and on morphological attributes, manly contribution area and slope. Contribution area maps represent important tools to define, from a DEM, the location of hollows in the field, helping in the prediction of saturated zones, inundation areas, drainage networks and landslides. In this work we calculate, from a 2 meters grid DEM, the contribution area maps of a 2 Km² basin using two different procedures. In the first, we developed an algorithm and used it in an inexpensive set of hardware and software, under a PC platform. In the second one, we used a robust software under a workstation environment. These procedures were applied to a basin in which many shallow landslides were triggered after intense rainfall in February 1996. The two procedures have presented different results. For example, only the second procedure was able to predict the location of hollows in the field, with good accuracy, the values of each pixel contribution area. Besides, this procedure was able to extract the drainage network, with high resolution, from a DEM. The results suggest that the second procedure may improve the performance of landslide hazard models.