Abstract: Rice is widely cultivated over a large climate spanin China, and thereby its growing environments vary greatly. Two main systems of rice cropping arethe double-season rice (early rice and late rice) and single-season rice (middle rice). In Norther China,only single-seasonrice is cultivated, whereas, inSouthern China,both can be cultivated due to the moderate climate. In different cultivars and environments, the ORYZA(v3) model has been widely used for rice growing simulation, while the model has been calibrated and validated in the world. The ORYZA(v3) model is the latest version updated from the ORYZA2000, by integrating new modules and routines to quantify daily dynamics of soil temperature, carbon, nitrogen, and environmental stresses.This model has been significantly improved with enhanced capability to simulate rice growth, development, and yield formation under non-stressed, water stressed, and nitrogen stressed conditions. Many studies have been conducted for the sensitivity and uncertainty analysis of parameters in the ORYZA model series. However, the temporal and spatial characteristics of parameter sensitivities in the model are still unclear. In this study, 18 typical sites were selected from 16 rice cultivation sub-regions in China, and a global sensitivity analysis for each site was conducted for 16 crop parameters in the ORYZA(v3) model over 30 years (1986-2015) using the Extended FAST method.Theoutput variables were set as the leaf area index (LAI), dry weight of stems (WST), total aboveground dry matter (WAGT), and dry weight of storage organs (WSO) at four development stages (the basic vegetative, photoperiod-sensitive, the panicle-formation, and grain-filling phase). The 30-year means and standard deviations of total sensitivity indices and interaction indices of each model parameters were calculated for the output variables at different growing stages under different environments. The correlations between parameter sensitivities and meteorological factors were analyzedto explore the impacts factors of model parameter sensitivities. The results showed that the total sensitivity indices and interaction indices greatly varied with different growing stages, different rice regimes, and different sites. Moreover, the differences were especially obvious in those sites, such as Dali, Yinchuan, and Mudanjiang,particularly onthe high altitude or special climate conditions, comparing to other typical sites. In the selected parameters, the sensitivities of parameters RGRLMX, SPGF and WGRMX were strongly influenced by the environment than others, thus their calibration need to be paid more attention. Correlation analysis indicated that the global sensitivity indices of model parameters wassimilar and significant correlations with daily maximum temperature, daily minimum temperature, and cumulative temperature within rice growing period. However, the sensitivity of model parameters cannot show a single pattern of variation in space,due to the integrated impacts of altitude, latitude and other factors on local climate conditions at the typical sites. In some typical sites, such as Dali and Mudanjiang, the RGRLMX has very high sensitivity, while the others were insensitive or little sensitive, indicating that the model may have some imperfections leading some limitation of applications in those areas with cold weather during rice growing period.