Using the static opaque chamber method, a field experiment was conducted in situ for two years to study the effects of three cultivation systems on CH4 and N2O emissions from permanently flooded rice fields in a hilly area in Southwest China. The results show that the average CH4 fluxes from a permanently flooded rice field with a single middle rice crop and flooded with no winter crop (PF) were (21.44 +/- 1.77) mg x (m2 x h)(-1) and (3.77 +/- 0.99) mg x (m2 x h)(-1) during rice-growing and non-rice growing periods, respectively, where both values were much lower than many previous reports from similar regions in Southwest China. The annual CH4 emission was mainly occurred in the rice growing period, being only 23.2% of the total annual CH4 flux emitted from the non-rice growing period, though the latter occupied two thirds of a year. The annual average flux of nitrous oxide was (0.051 +/- 0.008) mg x (m2 x h)(-1) and the N2O emission also intensive in the rice growing period. However, being only 8.1% of total annual N2O flux emitted from the non-rice growing period. After implementing the rice-wheat rotation (RW) and rice oil-seed rape rotation (RR), the CH4 emissions were reduced substantially, only 43.8% and 40.6% of those of PF, respectively. However, the N2O emissions were increased after conducting RW and RR systems, which were 3.7 and 4.5 times larger than those of PF. The global warming potentials (GWPs) of the CH4 and N2O emissions under the three tillage-cropping systems were assessed in an integrated way. The results show that the integrated GWPs of the CH4 and N2O emissions are in the following sequence: PF>>RR approximately equal to RR. Within 20, 100 and 500 years spans, the GWPs of the CH4 and N2O emissions of PF were 2.6, 2.1 and 1.7 times larger than those of RW (or RR), respectively. After introducing rice-wheat or rice oil-seed rape rotation systems into the permanently flooded rice fields, the integrated GWPs of the CH4 and N2O emissions were decreased largely.