Net ecosystem carbon exchange (NEE), an important indicator of the carbon cycle and carbon balance, is determined by gross primary productivity (GPP) and ecosystem respiration (ER). With the increased use of nitrogen (N) in industry and agriculture, N deposition has significantly increased globally. It has been widely documented that N enrichment stimulates plant growth, and thereby enhances productivity in grassland ecosystems. However, little is known about how N addition affects NEE and what are the differences in the conditions between various N addition levels. To address these questions, we conducted a manipulative experiment of N addition in a typical steppe in Inner Mongolia, China, in 2012 and 2013. Two treatments and a control (0 g N m^-2a^-1) were set up: a moderate (10 g N m^-2a^-1) and a high (40 g N m^-2a^-1) level of N addition. NEE was observed continuously in high time-resolution using a dynamic chamber system. Our results showed that the GPP of moderate N addition increased by 15.6% and 20% in 2012 and 2013, respectively, whereas there was no significant effect on ER. As a result of GPP enhancement, NEE significantly decreased (i.e., enhanced carbon sequestration capacity) by 230% and 337% in 2012 and 2013, respectively. In contrast, high N addition did not significantly affect GPP, ER, or NEE. Our results also showed that the enhanced carbon sequestration capacity was mainly attributable to increased soil water content and N concentration of plant leaves, and a decrease in soil temperature in the moderate N addition treatment. However, enhanced soil acidity and changes in species composition could have resulted in lower carbon sequestration capacities of the high N addition treatment. Our results indicated that the effects of N addition on GPP and ER depend on the amount of N added, which leads to remarkably different carbon sequestration capacities. The findings of this study could be used to improve process models for grassland ecosystems.