Abstract: Integration Regression is a new method of yield dynamic forecast because it has a clear biological significance and good effects on prediction.Adaptability of dynamic forecast to main crops was studied with integration regression method in Heilongjiang province for improving the method of dynamic forecast and making forecast accuracy better.The results indicate that the built crop yield dynamic forecast models for corn,rice and soybean based on integration regression method during late June to late September pass the test of F significance.The smaller mean difference (MD) and relative error (RE) prove that all the models have better performances.More specifically,the spring maize and rice prediction models have excellent performances and are able to reproduce practical yields well.With model testing on the basis of field experimental data from 2011 to 2014,spring maize,rice and soybean yield prediction models achieve respectively the simulation accuracies of 96%,95% and 93%,which manifests that the adaptability of integration regression method is slightly worse for soybean and better for spring maize and rice in Heilongjiang province.In another word,based on integration regression method,dynamic forecast for corn and rice is feasible in Heilongjiang province,while its adaptability for soybean will need to be further studied.