The present work aimed to develop mathematical model of grain filling of rice from primary and secondary branches within a panicle with increment to their weight (dry basis)-time after anthesis relationships. Three growth models for grain filling of rice were developed follow the exponential, the logistic and Gompertz functions. The grain filling models are expressed with parameters of mass at time zero, mass at time infinity and a measure for relative grain filling rate. Application of all models using data of thousand grain mass (dry basis) of rice from different branches within a panicle for Sintanur and IPB-4S variety were collected every 4 days during 14-30 days after anthesis. Model selection was conducted using Coefficient determination (R2), Root mean square error (RMSE) and Aikake's Information Criterion (AIC). R2, RMSE and AIC values for the Gompertz models were 0.999, 0.224, -9.949 (rice from primary branch for Sintanur); 0.997, 0.353, -4.512 (rice from secondary branch for Sintanur); 1.000, 0.131, -16.376 (rice from primary branch for IPB-4S) and 0.999, 0.266, -7.877 (rice from secondary branch for IPB-4S) respectively which revealed that theGompertz model was considered best to described the increment thousand grain mass (dry basis) of rice from different branches within a panicle for all two varieties.