Based on the composite diagnostic index, the relationship of meteorological factors (such as the daily average temperature and precipitation) between the forecasting year and any historical year during the growing season of American wheat is investigated, and the forecasting model based on the ground meteorological data is built by using the historical meteorological influence index. Meanwhile another forecasting model is built, using the monthly average sea surface temperature of the West Pacific Ocean and the circulation data at 500 hPa in the Northern Hemisphere as the main influencing factors. The ensemble forecasting model is built based on the stability of various models by the weighted method. The forecasting verification for 1995-2004 and experiments for 2005 and 2006 indicate that the accuracy of the ensemble model is over 92% and can meet the needs of operational service.