In order to make up shortage of existing forecasting models and improve forecasting precision and effectiveness, a new thought of combination forecasting model is given in this paper. Combination forecasting model based on drift is proposed by studying the drift and complementarity of different single forecasting models. The calculation steps of combination forecasting model based on drift are given. A practical example on Chinese energy consumption shows four forecast efficient measures (SSE, MAE, RMSE, MAPE) of combination forecasting model based on drift are better than single forecasting models, combination forecasting model based on drift can improve forecasting precision and it is effective in practice. Combination forecasting model based on drift is an effective forecasting model to make up shortage of existing forecasting models.