The precision of TRMM 3B43V7 satellite precipitation data were evaluated at different temporal and spatial scales based on the observed data from 45 meteorological stations in the Red River Basin and surrounding areas from 1998 to 2015. The impact of elevation, slope angle and slope aspect on accuracy of TRMM precipitation data were also analyzed. The results showed that: at the annual scale, the goodness of fit between TRMM estimates and the measured precipitation data was 0.75, and the former was 7.73% higher than the latter overall. Meanwhile, at the monthly scale, the goodness of fit between the two data was 0.84, the significant correlation was found between them. At the seasonal scale, the goodness of fit of precipitation (R2=0.77) in spring was higher than that in other three seasons, and goodness of fit for winter precipitation was the weakest (R2=0.64). At the catchment scale, precipitation focus movement showed that TRMM precipitation data can basically reflect the spatial distribution and evolution process of precipitation. At the site scale, the correlation coefficients of all stations were higher than 0.84 and stations with relatively large bias were mainly distributed in valley and basin areas. The accuracy of TRMM data was relatively high over the region where elevation was higher than 1000 m, slope was less than 2° and in the southeast direction. Principal component analysis showed that the influences of slope and aspect on the accuracy of TRMM data were greater than that of elevation.