Based on continuous three-year measurements (from 2004 to 2007) of eddy covariance and related environmental factors, environmental controls on variation in soil respiration (Rs) during non-growing season were explored in a maize agroecosystem in Northeast China. Our results indicated that during non-growing seasons, daily Rs was 1.08-4.08 g CO2 m-2 d-1, and the lowest occurred in late November. The average Rs of non-growing season was 456.06 ± 20.01 g CO2 m-2, accounting for 11% of the gross primary production (GPP) of the growing season. Additionally, at monthly scale, the lowest value of Rs appeared in January or February. From the beginning to the end of non-growing season, daily Rs tended to decrease first, and then increase to the highest. There was a significant quadratic curve relationship between Rs and soil temperature at 10 cm depth when soil temperature was more than 0°C (P<0.001), with the explaining ratio of 38%-70%. When soil water content was more than 0.1 m3 m-3, soil moisture at 10 cm depth was significantly parabolically correlated with Rs (P<0.001), explaining the rate of 18%-60%. Based on all the data of soil temperature of more than 0°C, a better model for Rs was established by coupling soil temperature and moisture, which could explain the rate of up to 53%-79%. Meanwhile, the standard error of regression estimation between the values of prediction and observation for Rs could reach 2.7%-11.8%. Rs in non-growing season can account for 22.4% of Rs in growing season, indicating that it plays a critical role in assessing the carbon budget in maize agroecosystem, Northeast China. Based on continuous three-year measurements (from 2004 to 2007) of eddy covariance and related environmental factors, environmental controls on variation in soil respiration (Rs) during non-growing season were explored in a maize agroecosystem in Northeast China. Our results indicated that from during non-growing seasons, daily Rs was 1.08-4.08 g CO2 m-2 d-1, and the final occurred in late November. The average Rs of non-growing season was 456.06 ± 20.01 g CO2 m-2, accounting for 11 % of the gross primary production (GPP) of the growing season. Additionally, at monthly scale, the lowest value of Rs since January or February. From the beginning to the end of non-growing season, daily Rs tended to decrease first, and then increase to the highest. There was a significant quadratic curve relationship between Rs and soil temperature at 10 cm depth when soil temperature was more than 0 ° C (P <0.001), with the explaining quadratic curve relationship between 38% -70%. soil water content was more than 0.1 m3 m-3, soi l moisture at 10 cm depth was significantly parabolically correlated with Rs (P <0.001), explaining the rate of 18% -60%. Based on all the data of soil temperature of more than 0 ° C, a better model for Rs was established by coupling soil temperature and moisture, which could explain the rate of up to 53% -79%. Meanwhile, the standard error of regression estimation between the values ​​of prediction and observation for Rs could reach 2.7% -11.8%. Rs in non- growing season can account for 22.4% of Rs in growing season, indicating that it plays a critical role in assessing the carbon budget in maize agroecosystem, Northeast China.