Using a static chamber-gas chromatography technique, we simultaneously measured the emissions of N2O and CH4 from paddy soils under three cropping systems (rice-fallow (RF), rice-wheat (RW) and maize-wheat (MW)) with three nitrogen (N) levels (N0, no N; Nopt, optimized N fertilization; and Ncon, conventional N fertilization) in a two-year field experiment in southwest China. Annual mean global warming potential (GWP) of CH4 and N2O was lowest (1775 to 2288 kg CO2-eq ha−1) in the RW system, medium in MW system (1915 to 6186 kg CO2-eq ha−1) and highest in RF system (7324 to 8112 kg CO2-eq ha−1). N2O emissions in N treatments accounted for approximately 90% of the GWP inMW system, and up to 40% of the GWP in RW system, and b8% of the GWP in RF system. The ranking of the annual mean yield-scaled GWP of CH4 and N2O was RW system (0.22–0.24 kg CO2-eq kg−1 grain) b MW system (0.36–0.96 kg CO2-eq kg−1 grain) b RF system (1.04–1.36 kg CO2-eq kg−1 grain). In the present study, the average N2O emission factor (EFd) was higher inWMsystem (Nopt: 1.82%, Ncon: 2.27%) than that in RWsystem (Nopt: 0.36%, Ncon: 0.55%) and in RF system (Nopt: 0.23%, Ncon: 0.37%). Given the comparable area- and yield-scaled GWP between the Nopt and Ncon treatments for RW and RF systems, the Nopt treatment that reduced local farmers' N fertilizer application rate by 33–40% maintained crop yields and decreased N2O emission and GWP of N2O plus CH4 in the subtropical rotation systems of southeast China.