Farmland ecosystems constitute great carbon sources in terrestrial ecosystems and have important influence on the global carbon cycle. As China has promised to reduce carbon emission significantly, China's agriculture is facing the challenge of carbon sequestration and reduction of carbon emission. As the environment in karst area of southwest China is unique and upland field covers a large scale, further research is badly needed to quantify diurnal and seasonal variations of soil CO₂ flux, and to investigate the effects of temperature, soil water content and soil properties on soil CO₂ flux. Taking upland field with corn-rapeseed rotation in Kaiyang country of Guizhou province as our study subject, we used static enclosed chamber coupled with GC to observe the soil CO₂ flux of all the rotation period. Continuous measurements of soil temperature and moisture were conducted together with several other physical and chemical properties of the soil. It was found that the upland with corn-rapeseed rotation performed as the source of CO₂. The mean CO₂ flux was (178.8±104.8) mg CO₂·m^-2·h^-1 in the rapeseed growing season, (403.0±178.8) mg CO₂·m^-2·h^-1 in the maize growing season, and (271.09±176.37) mg CO₂·m^-2·h^-1 for the whole rotation period. These values are in the upper part of the range of published soil-surface CO₂ flux date. There were significant positive correlations between soil CO₂ flux and temperature in both diurnal changes and seasonal fluctuations. The diurnal fluctuations of CO₂ flux showed similar patterns and fluctuated from minimum around 12 pm to maximum at 1 pm to 3 pm. For the seasonal variations, soil CO₂ flux was positively correlated with temperature in the whole crop rotation with the minimum value of 31.10 mg CO₂·m^-2·h^-1 in February and the maximum value of 665.46 mg CO₂·m^-2·h^-1 in August. An exponential function provided the best fit for describing the relationship between seasonal soil CO₂ flux and temperature for the whole crop rotation, with the higher correlation found with the soil temperature at 5 cm depth. About 86.7% of the seasonal soil CO₂ flux variations can be explained by air temperature, while 80.5% by the soil temperature at 5 cm depth. The results showed that the temperature was the determinant factor controlling temporal variation of soil CO₂ flux both diurnally and seasonally. The correlations between soil CO₂ flux and soil water content differed for the two kinds of crops. For the rapeseed growing season when soil water content was relatively high, the soil CO₂ flux and soil water content had a positive correlation. While there was a negative relationship between soil CO₂ flux and soil water content for the maize growing season. Both the relationships were not statistically significant which indicated that the soil water content was not the determinant factor for the soil CO₂ flux changes. The Q₁₀ value was calculated as 2.02 based on the atmospheric temperature, higher than pinewood at similar latitudes and evergreen broad-leaved forest at lower latitudes. This meant that the upland field corresponded more sensitively to the temperature changes. The significant positive correlation between soil CO₂ flux and soil pH indicated that the soil CO₂ flux was influenced by several parameters at the same time.