Abstract: Investigation of the progress in methods of agrometeorological disaster prediction in China in recent years is made, some new methods and techniques are introduced, and some comments are presented. It is concluded that: ① At present the statistical models are the major approaches to predict the agrometeorological disasters; recently some new methods and techniques are used, including improvement of statistical models, combination between climate models and agrometeorological models as well as application of crop growth simulation models. In general the research of agrometeorological disaster prediction is not very satisfied. ② Attention should be paid to determination of index, physical mechanism and agricultural meaning of predictors and processing methods of mathematics models, as well as prediction method based on crop simulation models. ③ It is the effective approach for agrometeorological disaster prediction to synthesize different disciplines, multi-prediction methods, multi-time scale of predictions and to combine prediction models with monitoring information by remote sensing. The operational early warning system of agrometeorological disaster prediction should be set up.