Ecosystem and climatic changes due to anthropogenic emissions of CO2 are of increasing importance to society. One way to predict these changes in the future is to study warming events in the geologic past. The main control of Earth's temperature over geologic time is the concentration of atmospheric CO2; however, the precise relationship between CO2 and temperature is not fully understood. Therefore, it is essential to be able to quantify atmospheric CO2 concentrations in the geologic past, especially at times of rapid climate change analogous to current atmospheric changes. One widely applied proxy relates the carbon isotopic composition of pedogenic carbonates to atmospheric pCO2, but one of the key variables (soil-respired CO2; S[z]) is not well constrained. This study presents a new proxy where soil-respired CO2 is related to mean annual precipitation (MAP) by the following equation: S(z) = 5.67(MAP) – 269.9, R2 = 0.59, SE = 681 ppm. This proxy, validated for modern atmospheric pCO2 levels, constrains the primary source of uncertainty in the soil carbonate paleobarometer. We apply this proxy to make atmospheric pCO2 reconstructions for examples from the Cenozoic, Mesozoic, and Paleozoic and calculate CO2 estimates that are in better agreement with estimates from other independent proxies and model results than previous pedogenic carbonate reconstructions. The implications of the uncertainties attributed to pedogenic carbonate paleobarometry as well as guidelines for using this method are discussed. By making individual measurements rather than assumptions for each isotopic value and combining those analyses with an S(z) estimate for each paleosol considered using our new proxy described herein, we can obtain increased precision for atmospheric pCO2 reconstructions.