Soil heterotrophic respiration is a major way leading to losses of soil carbon into the atmosphere and plays an important role in global carbon cycle. Global warming may cause increases in rainfall or droughts that would enhance the variation of soil moisture. However, it is unclear that how the soil respiration will respond the co-effects of the simultaneous changes in air temperature and soil moisture. Our experimental site was located in a wheat field in the Loess Plateau of China. Rainfall was the sole way to deliver water into the soil at the site. It was observed that three heavy rainfall events caused significant alterations of the soil moisture in the period from spring to summer in 2005. During the same period, the air temperature increased significantly due to the monsoon climate. Soil respiration rates were measured in situ with three chambers of an automated multi-channel chamber system; relevant environmental factors were also simultaneously recorded. Correlations of the soil respiration rates with (1) the air temperature, (2) the soil moisture, and (3) both the air temperature and soil moisture were calculated. Temperature dependent models, soil moisture dependent models and double predictor models which based on both air temperature and soil moisture were used to fit the data. Through the tests against our field data sets, we built a E-Q model as SR=aebT(c+dW+fW2)g. Our conclusions are as flows: (1) the single predictor models based on only air temperature were not capable of predicting the soil respiration rates for the experimental field due to the significant alterations in the soil moisture; (2) among the soil moisture dependent models, the quadratic model was better than the linear model or the exponential model; (3) the E-Q model, which predicted soil respiration rates based on the exponential relation with air temperature as well as the opposite effect of soil moisture, was more capable for soil respiration predictions for the fields in this climate zone.