Abstract: The paper aims to overcome previous wheat strip rust prediction during winter and spring which were mainly based on fixed ten-day or monthly climate factors, and improve the accuracy and timeliness of extended prediction. Figures of the prevalence area used in this study were derived from the weekly wheat stripe rust reports in Nanyang from 2008 to 2015. Two methods were adopted to establish the dynamic prediction model based on the prevalence stages and the prevalence rate, respectively. The prediction accuracy of the prediction model based on three prevalence stages was 90% for the beginning stage of the prevalence, 95% for the developing stage of the prevalence, and 94% for the peak period of the prevalence. The prediction accuracy of the prediction model based on prevalence rate was 70%, 80% and 90% for the prevalence area rate (r1), prevalence area plus controlled area rate (r2) and prevalence area plus half controlled area rate (r3), respectively. The prediction model of this research improved the accuracy and timeliness for wheat strip rust prediction.