Abstract: Combinating several forecast models of a single object together is an efficient way to improve the precision in forecast. Every forecast model of a single object in the combination is evaluated comprehensively, in views of experts' rational expectation, on its adequacy of fit models, method complexity in application, compatibility and time dependency. Then, a combination weight coefficient is set up so as to form a combination forecast method characterized by subjective information integrated with objective information. Comparing the combination model with the models of single object analytically, a forecast is made for the quantity of private cars in China with an analysis made to the forecast. The results reveal that the former is superior to the latter in adaptability with precision improved greatly.