A small-scale, controlled carbon dioxide (CO2) release experiment was conducted at the Brackenridge Field Laboratory, Austin, Texas, to simulate CO2 leakage from a geological sequestration site. The main purpose of the experiment was to assess impacts of soil physical properties on soil CO2 measurements for detecting near-surface CO2 leakage signals. The field site includes one shallow CO2 injection well (1.1-m depth) and three wells for monitoring soil CO2 concentrations. CO2 was released at a depth of 1.1 m below the surface for approximately 6 hours. CO2 concentrations at the sensor locations clearly showed the arrival of CO2 from the injection well. A numerical model accounting for CO2 diffusion and dissolution into soil pore water was constructed and then calibrated with the CO2 measurements from the experiment. Using the calibrated numerical model, a set of sensitivity runs was conducted to assess effects of soil physical properties on modeled CO2 concentration (or flux). Results of the sensitivity runs show that CO2 concentrations (or fluxes) are sensitive to change in soil physical properties. Peak CO2 concentrations modeled in the sensitivity runs are most sensitive to soil porosity, and peak CO2 fluxes are most sensitive to the soil impedance factor. This study indicates that soil physical properties can have significant impacts on the soil CO2 measurements used for detecting CO2 leakage signals.