Optimum harvest time and yield gap information are important aspects of grain quality optimization and production development. The World Food Studies (WOFOST) crop simulation model was studied in its application for soybean optimum harvest time and yield gap analysis in East Java, Indonesia. Data inputs were local weather of solar irradiance and daily temperature, with given soybean varieties provided in the WOFOST simulation. The simulation result was validated with the actual data using homogeneity test of regression coefficient. Result showed that differences between simulation and actual yield were insignificant (α=0.05), for each tested locations and soybean varieties. The average potential yield was 1,716 kg ha-1, where the highest was obtained from S-France 904 variety located in Malang Regency. The optimum root mean square error was 49.42 kg ha-1 with correlation coefficient of 0.918. Meanwhile, the optimum harvest time and yield gap have corresponded to the actual data where harvest time was at the shortest in Blitar Regency using N-France 901 and N Spain 903 varieties, while the average yield gap was 33%. In conclusion, WOFOST simulation model has a prospect to be applied further using local soybean varieties followed by validation in the whole East Java region.