A subsurface, gradient approach was developed to continuously monitor soil carbon dioxide fluxes. Carbon dioxide gas concentrations at 2, 10, and 18 cm beneath the soil surface were measured using 7.6 m lengths of expanded Teflon tubing that is porous to carbon dioxide and other soil gases. The tubing was placed into the soil using a soil-trenching machine. The soil gas that diffused into the inner tubing void was circulated in a closed-loop system every hour using an instrument suite that included a datalogger, pump, flowmeter, nondispersive-infrared carbon dioxide sensors, valves, pressure transducers, thermocouples, and manifolds. During circulation, the soil gas was analyzed for carbon dioxide concentrations. This process was repeated every hour for each depth. A Fick's law finite-difference approach was used to estimate the flux of carbon dioxide from the soil on an hourly basis. The diffusion coefficient of carbon dioxide through the soil was determined experimentally using a modification of the McIntyre and Philip (1964) approach. Soil carbon dioxide concentrations were determined to range from near ambient atmospheric levels at the 2 cm depth to about 15,000 ìmol mol 1 at the 18 cm depth and were greatest immediately following precipitation events. The diffusion coefficient of carbon dioxide in the soil was determined to be about 2.7 x 10-6 m2 s-1. Using Fick's law and the carbon dioxide concentrations from the 18 and 2 cm depths, the flux of carbon dioxide was estimated to range from about 0.25 to 4.5 ìmol m-2 s-1. This method addresses both the spatial and temporal variability that is inherent in soil systems by replication in time and spatial integration of measurements along the lengths of the Teflon tubing.