The stable carbon isotope composition of CO2 occluded in the gibbsite structure is proposed as a potential atmospheric paleo-PCO2 proxy. Analysis of pedogenic gibbsite from a modern Ultisol in the Piedmont of Georgia, USA, was conducted to test the basis for this concept and to help constrain the parameters used to describe physical and biological processes affecting such factors as the respiration rate of CO2. Co-variation of the δ13C and δ18O values with depth along a gradient parallel to the mixing line between the atmosphere and the soil organic material implies that diffusion is the process that determines the stable isotope composition of soil CO2. In the upper 40 cm, the measured δ13C values are not consistent with the expected diffusive depth profile assumed in paleo-PCO2 models. The isotope signature is reset downward in the depth profile with a concentration of the most atmosphere-like δ13C and δ18O values occurring at the top of the Bt horizon by some as-yet-unknown process. Bioturbation, recrystallization, and physical translocation are potential explanations for this observation. Regardless of the process at work, the net effect is an apparent two-component mixing curve between the top of the Bt horizon and deep within the saprolite. In cases where the A horizon is eroded but the Bt horizon is preserved it is possible that δ13C values of gibbsite-occluded CO2 can serve as a proxy for atmospheric paleo-PCO2. Careful textural study of all paleosols is therefore essential to match stable carbon isotope signatures with the horizons preserved. Understanding of modern dynamics and preservation of these isotopic signatures may also be important for those that employ other carbonate proxies.