Abstract: The classification scheme developed by Jenkinson and Collison (1977) based on a typing scheme of Lamb(1950) is applied to obtain circulation types from the mean sea-level pressure of Northeast of China on a monthly basis. Monthly mean sealevel pressure data from 1951 to 2002 is used to derive six circulation indices and to provide a circulation catalogue with 27 circulation types. Five major types (N, NW, C, CSW, SW) which occurred most frequently are analyzed to reveal their relationships with the temperature of Harbin on various time scales. Stepwise multiple regression is used to reconstruct temperature anomaly. The monthly mean rainfall of all types occurred and the composite maps of the three major types(C, CSW, SW) relevant to Harbin's precipitation are studied. The results show that the dominant types in winter are N and NW types. C, CSW and SW types occur frequently in summer. N and NW types favour a negative temperature anomaly and correspond to less rainfall, while C, CSW and SW types often induce a positive temperature anomaly and correspond to more rainfall. Moreover, a successful statistical model can be established with only one of the six indices and large-scale mean temperature. Using the model, 77.3% of the total variance in the temperature anomaly between 1951 and 2002 can be reconstructed. C type has a close relationship with total rainfall and C type precipitation played a major role in determining the total rainfall of Harbin in recent years. This classification scheme is a statistical downscaling model and its relationships with temperature and precipitation can be used to forecast the regional climate.