The use of oilseed rape (OS, Brassica napus L.) as a winter green manure is crucial for enhancing soil fertility and reducing chemical N application in paddy fields. However, the impacts of replacing varying amounts of chemical N with OS on CH4 and N2O emissions in paddy soils have not been well evaluated. In this study, GHG emissions, soil properties and OS decomposition in a rice-ratooning system with different OS-urea N replacement rates (0%, 25%, 50%, 75% and 100%) were investigated. Our results indicate that 84.7–90.7% of the initial C and 97.5–98.4% of the N were released during the 192-day decomposition process, and that the mineralization patterns of net C and net N in the OS residue were consistent with a single exponential decay model. The lowest CH4 emissions (9.97 g m−2) were observed at 0% OS, while the highest N2O emissions (0.40 g m−2) were observed at this level of substitution. Conversely, the highest CH4 emissions (20.71 g m−2) and lowest N2O emissions (0.07 g m−2) were observed at 100% OS. Compared to 0% substitution, 25% substitution significantly decreased GWP and GHGI without reducing rice grain yield. Environmental parameters such as soil redox, NH4+-N and residual N and C were shown to be significantly associated with CH4 emissions, whereas soil redox, NH4+-N and residual C were the main drivers of N2O emissions. In conclusion, 25% substitution of OS was the most cost-effective measure for balancing greenhouse gas emission and rice yield.