Abstract: The nine evaluation criteria of forecasting effects are introduced.And a combined forecasting approach based on multi-criteria optimization is proposed.This approach is different from the traditional combined forecasting approach by improving single criterion.By multi-criteria optimization and selecting some criteria which the effect of combined forecasting is superior to that of the single criterion,a nonlinear programming model is constructed.The weights of combined forecasting are obtained by using genetic algorithms to solve the nonlinear programming model.Finally an example is given to show the feasibility and availability of this method.