Regional wind power prediction is an effective way to increase the wind power penetration and improve the security and economy of the power system.Based on statistical upscaling approach，a regional wind power forecasting method is presented.In this method，the data of several representative wind farms are used to calculate regional results.Firstly，representative wind farms are selected not only based on the correlation between single farm and regional power，but also on the forecasting accuracy of each farm.Secondly，parameters are determined for each representative wind farm.Operation data of North China power grid from January to July，2011is used to validate the upscaling method.The result shows that the upscaling method is appropriate for the regional wind power forecast model.Compared with the conventional accumulating method，the statistical upscaling method will make the regional forecast model less dependent on the data integrity of each single farm，as well as improve the accuracy of forecast result.