Exploring optimal mulching methods for maize production is critical to achieving a balance between maximizing yield and reducing negative environmental impacts in dryland agriculture. Therefore, a 3-yr field experiment was conducted to investigate the effects of three mulching methods—biodegradable plastic film mulching (BM), maize straw mulching (SM), and no mulching (NM)—on spring maize yield, greenhouse gas (GHGs) emissions (CO2, CH4, and N2O), soil factors, the net ecosystem GHG balance and greenhouse gas emission intensity (GHGI) on the Loess Plateau of China. The result showed indicated that CO2 and N2O are the primary sources of GHGs in dryland maize production. The yield of BM significantly increased 1.43–1.99 t ha−1 compared to NM, CO2 emissions (on average 32.83%) compared to the NM. And BM showed suitability for N2O emissions (3.7–11.2%). SM increased CO2 and N2O emissions (by 9.43–13.17% and 24.96–33.64%, respectively) compared to the NM. Structural equation model analysis showed that soil temperature was the most direct factor to affect GHGs. SM was the sink of net ecosystem GHGs, BM and NM were the source of net ecosystem GHGs. These results indicated that mulching emerges as a promising strategy for dryland agriculture. BM method can be used when the primary goal was to improve yield, while SM method is a better choice when the primary goal was to improve soil organic carbon sequestration and reduce net GHGs emissions. This study clarified the impacts of mulching methods on three GHGs emissions to provide information support for sustainable agriculture on the Loess Plateau.