The initial fields and output of Mesoscale Numerical Prediction System in Northeast China were verified by means of WMO's standard verification method because there were some failures in the operational prediction There are two kinds of initial fields and output, one is that the initial fields were created by output of medium-range forecast model (T106L19) used as initial first guess fields and were corrected by sounding data set, called T106 The other is of the objectively analyzing fields of sounding data set, called OBS The results of verification as the follows: (1) T106 is better than OBS in the elements that represents synoptic scale, such as 300hPa's wind and 500hPa's height (2) OBS is better in the elements that represents mesoscale, such as 850hPa's temperature and humidity The output at the low level are more precious than that at the high level (3) The predicted winds at the high level are better, but there are some prediction errors in U and V component field At one time, TS skill scores of precipitation were calculated for two kinds of output These results indicate that TS score of T106 is higher than OBS's in general, but there are more rain of prediction with T106 initial fields with OBS's initial fields than observation rains.