A method of rainfall estimation by means of artificial neural network with the reflectivity, horizontal gradient and vertical integration of liquid water content is introduced, based on the radar data from the Hefei radar station in July 2007 and the Guangzhou radar station from May to October 2008 corresponding to the rain gauge data. The estimation of rainfall by the method is compared with the result of the 〖WTBX〗Z=300R1.4〖WTBZ〗 relationship. The results show that the rainfall estimation of the multi factor artificial neural network is better than that of the 〖WTBX〗Z=300R1.4〖WTBZ〗 relationship, according to the correlation coefficients and root mean square errors.