Abstract: The main problems of the multi-sources rainfall data quality evaluation consist of different time interval,different spatial characteristics and unknown tnie rain value.The multi-sources rainfall data can be classed into two groups:point rainfall and continuous-net rainfall.Based on the hypothesis that surface rain gauge data would be the true rainfall at one point and rader measured rainfall data corrected by gauge datawould reflect the spatial distribution of true rainfall.This paper gives five aspects of quality evaluation technique of multi-sources rainfall data:single point,synthetical points,mean area rainfall,spatial distribution,and products of flood forecast system.