Long-term agronomic studies are useful to determine cropping system nitrogen (N) use efficiency and the fate of applied fertilizers. We used a subtractive fertilizer experiment incorporating N, phosphorous (P), potassium (K) and swine manure to determine long-term changes in grain yield, soil organic carbon (SOC), total soil nitrogen (N), as well as carbon dioxide (CO2) and nitrous oxide (N2O) emissions. The experiment was conducted on a 22-year maize-soybean-wheat rotation in Northeastern China. Crop residues were removed for cooking fuel and forage according to local practices. Five fertilizer treatments were applied annually: control (no fertilizer), NK, NP, NPK, and NPKOM (N, P. K and manure). The NPKOM treatment increased SOC and total soil N by 4.59 and 0.45 Mg ha(-1), respectively. In contrast, SOC decreased by 10.6 and 6.64 Mg ha(-1) in the control and NK treatments, respectively. The NPKOM treatment had an average of 2.9 times more N2O emissions than the other fertilizer treatments. The cropping system balances for N and SOC, together with fuel use for farming practices and manure handling, were used to calculate the global warming potential (GWP) of the different fertilizer treatments. Due to SOC sequestration, the GWP of the NPKOM treatment (6.77 Mg C equivalent ha(-1)) was significantly lower than that of both the control (14.4 Mg C equivalent ha(-1)) and the NK treatment (12.8 Mg C equivalent ha(-1)). The results suggest that in rainfed agricultural systems in Northeastern China, the application of manure supplemented with NPK can simultaneously achieve higher grain yield and lower GWP compared to mineral fertilizers alone.