The problem of no match on time scale between single-time factors and predictor in objective forecast is analyzed, and the necessity and the possibility of using multi-time factors are put forward. Contrast experiment shows that multi-time factors can improve the quality of prediction equations and make effective rate reach 90.0 percents and reduce sum of residual squares 23.8 percents meanly. The reduction appears some regularity in spite of it's variation in various situation. It is also indicated that it is not benefit to contain more time number in multi-time factors in order to avoid “dimension disaster”.