In order to achieve maximum yield and seed quality, finding the best model to quantify 1000 seed weight changes at seed filling period and maturity is very important. This study was conducted to evaluate nonlinear regressions models for description wheat 1000- seed weight variation (cv. Zarrin and Sardari and Genotyps: C-81-10 and C83-8) response to time from flowering to harvest. Therefore, 1000- seed weight from different cultivars and genotypes of wheat was measured every seven days  in times of 20, 27, 34, 41, 48, 55, 62, 69, 76 and 83 days from flowering. Six functions (Segmented, Logestic, Vannella, Gompertz, Beta 1(original) and Beta 2 (Modified)) were evaluated to describe 1000- seed weight variation in response of time. The results indicated that  high RMSE ,low R²  and high CV observed in Segmented and Vannella models and the best model was Dent-like with optimum RMSE and R² and appropriate distribution around one by one line. In the all of cultivars and genotypes achieving to highest 1000-seed weight occurred in 65-67 days from flowering, thus it can be considered as physiological maturity time. In this time cv. Sardari and Zarrin had the highest and the lowest 1000-seed weight respectively. Generally results showed that based on Beta 2 model, in cultivars with higher 1000-seed weight, time of achieving to maximum rate of seed filling occurred earlier than others. Thus it could be recommended that the breeding and genetically programming make the seed filling time happen earlier but make the duration of seed filling longer.