Since the early 1980s, water management of rice paddies in China has changed substantially, with midseason drainage gradually replacing continuous flooding. This has provided an opportunity to estimate how a management alternative impacts greenhouse gas emissions at a large regional scale. We integrated a process-based model, DNDC, with a GIS database of paddy area, soil properties, and management factors. We simulated soil carbon sequestration (or net CO2 emission) and CH4 and N2O emissions from China's rice paddies (30 million ha), based on 1990 climate and management conditions, with two water management scenarios: continuous flooding and midseason drainage. The results indicated that this change in water management has reduced aggregate CH4 emissions about 40%, or 5 Tg CH4 yr-1, roughly 5--10% of total global methane emissions from rice paddies. The mitigating effect of midseason drainage on CH4 flux was highly uneven across the country; the highest flux reductions (>200 kg CH4-C ha-1 yr-1) were in Hainan, Sichuan, Hubei, and Guangdong provinces, with warmer weather and multiple-cropping rice systems. The smallest flux reductions (8.0 kg N/ha) in Jilin, Liaoning, Heilongjiang, and Xinjiang provinces, where the paddy soils contained relatively high organic matter. Shifting water management from continuous flooding to midseason drainage reduced total net CO2 emissions by 0.65 Tg CO2-C yr-1, which made a relatively small contribution to the net climate impact due to the low radiative potential of CO2. The change in water management had very different effects on net greenhouse gas mitigation when implemented across climatic zones, soil types, or cropping systems. Maximum CH4 reductions and minimum N2O increases were obtained when the mid-season draining was applied to rice paddies with warm weather, high soil clay content, and low soil organic matter content, for example, Sichuan, Hubei, Hunan, Guangdong, Guangxi, Anhui, and Jiangsu provinces, which have 60% of China's rice paddies and produce 65% of China's rice harvest.