Merging multi-information is a very important developing direction about hydrology modeling and prediction these days. Thereby, TRMM 3B42 data and rain gauges are adopted to engage in accuracy validation in Sichuan Basin and the surrounding areas as the study area because of frequent occurrence of devastating floods in the Yangtze River basin and this paper refers to correlation, spatial distribution, and bias due to elevation, based on single ground station rainfall and subbasin area precipitation. The results indicate that the daily and monthly correlation coefficient between TRMM 3B42 data and rain gauges rainfall is quite high (R>0.5), but it is low (R<0.5) for 3-h scale. There are significant deviations about 3-hourly and daily scales. Precipitation spatial distribution has good consistency between these two group data for different temporal scales. Probability density percentage of mean rainfall time series bias is 32.26% ± 10% from 2008 to 2010. TRMM 3B42 data are overestimated in most regions of Sichuan basin in the mass, but underestimated in the partially surrounding areas of Sichuan basin. The detection accuracy of 3-h scales TRMM 3B42 data is significantly lower than that of daily scale, and it is influenced to a certain extent by the elevation changes in the whole region. Because 3-h and daily scales of TRMM 3B42 data have no good precision, they need to be calibrated by the rain gauges, in order to use these data in hydrology modeling and prediction research.