Abstract: The aim of this study is to research the effects of prior information differences in regions and tree species on allometric biomass equation fitting, revealing the regularity which accurately select the prior information to improve the prediction of standing wood biomass. The mean and covariance matrix of parameters a and b are collected as prior information, coming from three regions of tropical, temperate, boreal and six genus of Quercus, Betula, Populus, Acer, Eucalyptus, Pinus and the above ground biomass data of Larix kaempferi is used to fit the allometric biomass equation. The data, based on the non-repetitive sampling 1 000 times is fitted by using the different prior information of Bayesian method. The regional and tree species difference of prior information had no significant influence on fitting effect of allometric biomass equation. However, it had significant influence on the predictive effect (P<0.01). MB and MRMSE evaluated from the total data and temperate are better than those of boreal and tropical, proving that regional factor affect the estimation accuracy of the allometric biomass equation and the prior information of selecting the region where the tree growth improves. In the process of tree species fitting, the predictive as prior information is Pinus and Quercus > the total data and Eucalyptus > Betula> Populus > Acer, indicating that tree species are the biological characteristics affecting the prior information. However, the biological relationship between Larix and Pinus > Larix and Eucalyptus > Larix and Acer > Larix and Populus > Larix and Quercus > Larix and Betula which are not consistent with the biological relationship between tree species. When using the Bayesian method to fit the allometric biomass equation, both environmental and genetic factors will improve the predictive accuracy of the model.