Abstract: The monitoring and information extraction of paddy fields using remote sensing techniques is an important means for modern agricultural management. However, it is difficult to obtain effective optical monitoring data of south China due to the frequent cloudy and rainy weather in spring and summer in this area. To accurately extract information on paddy fields in areas subject to frequent cloudy and rainy weather, this study investigated the paddy fields in Jiangxiang Town in Nanchang County, Jiangxi Province, using multi-temporal Sentinel-1 SAR data as the data source. Specifically, this study calculated the J-M distance between paddy fields and other land types in different phenological periods, analyzed the changes in the distance based on the backscattering coefficients of various land types in key phenological periods, and then obtained the best phenological images for the information extraction of paddy fields. Afterward, this study conducted ground object classification using methods such as random forest, maximum likelihood, support vector machine, and neural network and then compared and verified the classification accuracy. The results are as follows. The combined SAR data of the different stages including booting stage (June 14), trefeil stage (April 21), transplantion period (May 3), and transplanting peried of second season late rice (July 26) is the optimal temporal combination for the information extraction of paddy fields. Higher classification accuracy of ground objects in the study area can be obtained using the random forest method, with overall classification accuracy of up to 0.943 and a Kappa coefficient of 0.932. This study conducted the mapping of paddy fields in areas with frequent cloudy and rainy weather using SAR data and will provide important references for the temporal selection and classification.