Aims Balancing yield and soil CO2 emissions in cropping systems has become a focus of agricultural research. To explore effective ways to achieve the balance, three years (2016-2018) of field experiments were conducted in the oasis zone of Northwest China. Methods Three replications of the four tillage systems (NT, no-tillage; MT, minimal-tillage; ST, sub-tillage; FT, fold-tillage) were used in a randomized complete block design. Results It was found that tillage practice had no significant effect on soil carbon pools. However, FT exhibited a significantly higher soil carbon emission intensity (SCEI) and net ecosystem primary productivity (NPP). Still, poorly in terms of net ecosystem productivity (NEP) than those of the NT, MT, and ST treatments, MT performed the highest NEP and maize yield (P < 0.05). The total cumulative CO2 emissions of the FT were 23.3%-37.6% higher than those of the other tillage practices. Furthermore, we identified that soil CO2 emission was mainly influenced by soil temperature and soil moisture. NT increased farmers' net income and output/input ratio by 20.2% and 41.3% respectively compared with FT and demonstrated better economic benefits and net ecosystem productivity. Conclusions The results suggest that conservation tillage (especially NT and MT) is advantageous to local economic and ecological benefits in the short term. However, the years of continuous tillage are critical for increasing crop yields and reducing CO2 emissions, so the feedback response of long-term tillage on farmland still needs to be observed and studied.