In this paper, the basic ideas of combination forecasting and the current situation of studies on the Bayesian models of combining forecasts are reviewed. One of the reasearch procaduLres is extended to the case of multi-step Bayesian updation of information, and two Bayesian models of unbiased combining forecasts are conducted under different assumptions of characteristics of individual forecast bias. Furthermore, a Bayesian model of combining weights is developed with non-sample and sample information being combined ．Finally，areas of further research are pointed out