Currently, limited information is available regarding the effects of different farmland management practices on fluxes of methane (CH₄) and nitrous oxide (N₂O) from the fields. In this study, diurnal and seasonal variations of CH₄ and N₂O fluxes from fields under different farming practices, i.e. nowadays commonly used mulched drip irrigation (MD) and traditional mulch-free flood-irrigation (MFF), were monitored with closed static chambers and a modified gas chromatograph (HP4890). This study was conducted in a cotton field in Xinjiang, Northwestern China. Results show that the diurnal variations of CH₄ in treatments MFF and MD were in sharp contrast. In the former CH₄ flux descended first and then ascended, while in the latter it proceeded reversely. Over the whole cotton growing season, however, it had no clear pattern to follow. The soil in the former absorbed a total of CH₄ 0.7 ~ 23.1 mg m^-2a^-1 from the atmosphere, while the soil in the latter released a total of CH₄ 45.2 ~ 52.5 mg m^-2a^-1. As for N₂O fluxes, the two treatments exhibited similar diurnal and seasonal patterns of a curve going up first and down later. However, in treatment MFF daily N₂O flux was much higher than in treatment MD. Over the whole cotton growing season, the total N₂O emission was higher in treatment MFF (N₂O 99.3 ~ 320.0 mg m^-2a^-1 ) than in treatment MD (N₂O 60.0 ~ 259.0 mg m^-2a^-1). All the findings suggest that the MD system could alter the direction of the CH₄ flux between the soil and the atmosphere of field agro-ecosystem in the arid region. It enhanced considerably CH₄ emission, but did not affect much the diurnal or seasonal variation of N₂O fluxes, however, it did reduce N₂O emission from the soil quite significantly.