HARVEST INDEX is one of the important parameters in estimation of modules of seed yield prediction. Since, this parameter undergoes changes under various conditions. Therefore, this research was aimed to prepare relations to estimate it. for this purpuse, a compound analysis experiment with randomized complete block design with four replications was conducted at the Gorgan University of Agricultural Sciences and Natural Resources Univercity Research Farm in 2005. The treatments were four wheat varieties (Koohdasht, Shirudi, Tajan, Zagros) and six different sowing dates (14 December, 20 Jan, 20 Feb, 20 March, 16 Apr, 17 May). The results indicated that a significant difference was obvious in different sawing dates in respect to the slope of HARVEST INDEX line (dHI/dt). These changes were attributed to mean temperature, pre-Grain filling accumulated biomass, and fraction of post-anthesis (fG) producted dry matter. Therefore, based on obtained results and evaluated models, fG was the best optian for simulation of dHI/dt and HARVEST INDEX, in this respect, RMSE of dHI/dt for all varities by fG was 0.25 percentage per day which was the lowest one. Therefore, these relations can be used to evaluate the slope of HARVEST INDEX line, or to model grain yield in crop simulation models.