The performance of the Tropical Rainfall Measuring Mission (TRMM) rainfall algorithms was investigated from December 1997 to November 2000 over South America. The Precipitation Radar (PR) and the TRMM Microwave Imager (TMI) algorithms are compared with the rainfall estimates of the Geostationary Environmental Satellite (GOES) Precipitation Index (GPI), the Global Precipitation Climatology Project (GPCP), the gauge analysis of GPCP’s Global Precipitation Climatology Center (GPCC), and the gauge dataset from the Brazilian Agency for Energy and Electricity (ANEEL). Despite the good agreement of all rainfall estimates for long-term averages in the regions with high density of rain gauges from ANEEL, some discrepancies in the satellite estimates are observed comparing with the gauge estimates. Seasonal and regional analyses showed high estimates of PR, TMI, and GPI during the September-November season, especially in the regions of a large occurrence of Mesoscale Convective Systems (MCSs). Several explanations for large differences between satellite and gauges estimates were considered. The first could be related to sampling problems in the satellite estimates if the PR algorithm is overestimating rainfall from large precipitation features, because frequent occurrence of MCSs in some regions in South America appeared to be related to high rainfall estimated by the PR. The high estimates of GPI in some grids could be related to overestimation of rainfall from high clouds.