The study carried out different treatments consisted of different numbers of duck in the rice-duck ecosystem compared with the no-duck rice farming in an early and late season. The treatments were designed with triplication by randomized blocks. Methane emission fluxes from paddy soils were measured by the static chambers technique. The correlations between methane emission and soil physical and chemical characteristics were also analyzed. The results showed significant differences (p < 0.01) existed in the dissolved oxygen contents of water in the treatment fields, and that the more ducks, the higher the dissolved oxygen content. The dissolved oxygen content of 20 ducks per 667m2 was higher than any other situation compared with no-duck rice farming. The oxygen dissolve increased by 2.2%-68.7% and 11.07%-110.77% in the early and late season respectively. Secondly, the soil redox matter content and methanogenic bacteria population of the rice-duck ecosystem reduced more sharply than the no-duck rice farming resulting in a lower methane production. Moreover, methane emission fluxes at different growth stages varied as the methanogenic bacteria population varied. Thirdly, the amount of methane emission differed between the treatments - the more ducks, the less methane emission. Methane emission of rice-duck ecosystem reduced by 18.22%-28.13% during the early season and by 17.73%-34.44% during the late season in comparison with no-duck rice farming. The other related analysis showed that the negative correlation was significant (p＜0.001) between the methane emission flux and dissolution oxygen content of water body. However, CH4 emission flux had significant positive correlation (p＜0.01) with the soil redox matter content and rice field methanogenic population. Therefore, more ducks in the rice-duck ecosystem resulted in lower methane emission than the no-duck rice farming. The optimal number of duck has great effect on methane mitigation.