Mercury emission from natural resources contributes greatly to global atmospheric mercury, thus having an important impact on circulation of atmospheric mercury. Mercury emission during the earth surface processes is a major natural source of mercury in the atmosphere. In view of the shortcomings of the current methods for determining mercury concentration in soil air, this study has developed a new method. To test the method, soil air was collected from profiles of paddy soils in the Nanjing Liuhe Circular Agriculture Ecological Zone for analysis of total mercury. Using the experimental device, an inverted funnel, soil air in the soil profile was pumped continuously at a low flow rate, into a gold-coated pipe for pre-enrichment of mercury in the soil air. Gold-coated quartz sands were used as adsorbent to collect gaseous mercury in the soil air and the adsorption process lasted 3 hours with adsorption efficiency reaching nearly as high as 100% and relative standard deviation being 2.4%~5.4%. The highest mercury concentration was detected in the soil air extracted from the topsoil layer (0~3 cm) and the concentration decreased significantly with soil depth, but leveled off after the depth went beyond 20 cm. The experiment on effect of sampling flow on accuracy of the measurement, shows that when the sampling flow rate was below 30 ml min^-1 RSD of the measurement was <10.0% and when the sampling flow rate went beyond 30 ml min^-1, RSD increased, which indicates that at a higher flow rate than 30 ml min^-1, the device may suck some air from the atmosphere into its chamber, and a flow rate of 20 ml min^-1 is a safe one that enables the device to extract soil air merely from the soil profile. Then the air samples were analyzed with the cold vapor atomic fluorescence (CVAFS) method. Results show that the absolute detection limit is 0.023 ng m^-3. Air mercury concentration in the paddy soils varied in range from 6 to 18.94 ng m^-3. When the parallel experimental device was used to determine mercury concentrations of the air in the laboratory and soil air in the farmland simultaneously relative standard deviations of two measurements were both <15%. The comparison experiments show that the device collects air samples merely from soil profiles, rather than from the atmosphere above the soil surface. Mercury concentration of the soil air in paddy soils peaked at noon, which may be attributed to the higher temperature in the topsoil, intensive light and effective radiation during the noon time, enhancing photochemical reactions of mercury and increasing mercury concentration in the soil air. The highest concentration of gaseous mercury in the soil air was detected in soils at 6 cm in depth of the soil profile and then in soils at 3 cm in depth, which suggests that mercury in the soil air of the topsoil escapes into the atmosphere rapidly and its diffusion at 6 cm is retarded by soil. Additionally, the higher water content at 6 cm than at 3 cm may provide profitable additions for mercury in soil and soil water to convert into Hg0 in soil air. The soil air in underlying soil layers is relatively stable and less changed, which further proves the method is reliable. This method has the following advantages: during the experiment, lithium batteries power the device, which is easy to operate in the field and capable of collecting soil air at different depths, and enables spatio-temporal synchronization of observation of mercury concentrations in the soil profile. But it should be noted that this experiment can only be carried out in paddy fields unsaturated with soil water and the use of rotameter may lead to errors in flow measurement. This experiment is characterized by simplicity of the devices, and easy operation in field and can be used to precisely and accurately measure gaseous mercury concentrations in soil air in unsaturated paddy fields.