Soil CO2 emission from an arable soil was measured by closed chamber method to quantify year-round soil flux and to develop an equation to predict flux using soil temperature, dissolved organic carbon(DOC) and soil moisture content. Soil CO2 flux, soil temperature, DOC and soil moisture content were determined on selected days during the experiment from August 1999 to July 2000, at the Ecological Station of Red Soil, the Chinese Academy of Sciences, in a subtropical region of China. Soil CO2 fluxes were generally higher in summer and autumn than in winter and spring, and had a seasonal patt more similar to soil temperature and DOC than soil moisture. The estimation was 2.23 kgCO2/(m2·a) for average annual soil CO2 flux. Regressed separately, the reasons for soil flux variability were 86.6% from soil temperature, 58.8% from DOC, and 26.3% from soil moisture, respectively. Regressed jointly, a multiple equation was developed by the above three variables that explained approximately 85.2% of the flux variance, however by stepwise regression, soil temperature was the dominant affecting soil flux. Based on the exponential equation developed from soil temperature, the predicted annual flux was 2.49 kgCO2/(m2·a), and essentially equal to the measured one. It is suggested the exponential relationship between soil flux and soil temperature could be used for accurately predicting soil CO2 flux from arable soil in subtropical regions of China.