A long-term fertilization experiment was set up in northern Xinjiang, China to evaluate the dynamics of crop production and soil organic carbon （SOC） from 1990 to 2012 with seven fertilization treatments. The seven treatments included an unfertilized control （CK） and six different combinations of phosphorus （P）, potassium （K）, nitrogen （N）, straw （S） and animal manure （M）. The balanced fertilization treatments had significantly （P〈0.05） higher average yields than the unbalanced ones. The treatment with 2/3 N from potassium sulfate （NPK） and 1/3 N from farmyard manure （NPKM） had a higher average yield than the other treatments. The average yields （over the 23 years） in the treatments of NPK, and urea, calcium superphosphate （NP） did not differ significantly （P〉0.05） but were higher than that in the treatment with urea and potassium sulfate （NK; P〈0.05）. The results also show that the highest increases in SOC （P〈0.05） occurred in NPKM with a potential increase of 1.2 t C/（hm2.a）. The increase in SOC was only 0.31, 0.30 and 0.12 t C/（hm2.a） for NPKS （9/10 N from NPK and 1/10 N from straw）, NPK and NP, respectively; and the SOC in the NP, NK and CK treatments were approaching equilibrium and so did not rise or fall significantly over the 23-year experiment. A complete NPK plus manure fertilization program is recommended for this extremely arid region to maximize both yields and carbon sequestration.
