Abstract: We used the membership function theory of fuzzy mathematics to set up appropriate index model and compositive meteorological appropriate index model, these models were established by using ten days light-temperature-water meteorological elements of rice growth stages of representative points of different rice region in Sichuan Basin, and using integral regression and correlation coefficient method to determine the weight coefficient, which was determined by weather appropriate index at different growth stages of rice and meteorological yield of rice.Then we set up rice dynamic weather appropriate index model of the representative points from transplanting to maturity.Finally, we established regression forecasting model by using calculated value of dynamic weather appropriate index model and relative meteorological yield of rice to predict trend of weather yield of rice.The results showed the dynamic prediction model of rice meteorological yield passed significant testing, and it can dynamic forecast change tendency of meteorological yield after the period from heading to flowering of rice.