To upscale the genetic parameters of CERES-Rice in regional applications, Jiangsu Province, the second largest rice producing province in China, was taken as an example. The province was divided into four rice regions with different rice variety types, and five to six sites in each region were selected. Then the eight genetic parameters of CERES-Rice, particularly the four parameters related to the yield, were modified and validated using the Trial and Error Method and the local statistical data of rice yield at a county level from 2001 to 2004, combined with the regional experiments of rice varieties in the province as well as the local meteorological and soil data (Method 1). The simulated results of Method 1 were compared with those of other three traditional methods upscaling the genetic parameters, i.e., using one-site experimental data from a local representative rice variety (Method 2), using local long-term rice yield data at a county level after deducting the trend yield due to progress of science and technology (Method 3), and using rice yield data at a super scale, such as provincial, ecological zone, country or continent levels (Method 4). The results showed that the best fitness was obtained by using the Method 1. The coefficients of correlation between the simulated yield and the statistical yield in the Method 1 were significant at 0.05 or 0.01 levels and the root mean squared error (RMSE) values were less than 9% for all the four rice regions. The method for upscaling the genetic parameters of CERES-Rice presented is not only valuable for the impact studies of climate change, but also favorable to provide a methodology for reference in crop model applications to the other regional studies. To upscale the genetic parameters of CERES-Rice in regional applications, Jiangsu Province, the second largest rice producing province in China, was taken as an example. The province was divided into four rice regions with different rice variety types, and five to six sites Then the eight genetic parameters of CERES-Rice, particularly the four parameters related to the yield, were modified and validated using the Trial and Error Method and the local statistical data of rice yield at a county level from 2001 to 2004, combined with the regional experiments of rice varieties in the province as well as the local meteorological and soil data (Method 1). The simulated results of Method 1 were compared with those of the other three traditional methods upscaling the genetic parameters, ie, using one-site experimental data from a local representative rice variety (Method 2), using local long-term rice yield data at a county level after deducting the trend yield due t o progress of science and technology (Method 3), and using rice yield data at a super scale, such as provincial, ecological zone, country or continent levels (Method 4). The results showed that the best fitness was obtained by using the Method 1. The coefficients of correlation between the simulated yield and the statistical yield in the the method 1 were significant at 0.05 or 0.01 levels and the root mean squared error (RMSE) values ​​were less than 9% for all the four rice regions. The method for upscaling the genetic parameters of CERES-Rice presented is not only valuable for the impact studies of climate change, but also favorable to provide a methodology for reference in crop model applications to the other regional studies.