Carbon isotope analysis of paleosol carbonates is one of the most widely used methods for producing quantitative estimates of CO2 levels in the ancient atmosphere, and is increasingly used to estimate soil pCO2 as a proxy for primary productivity. Recent efforts to improve the carbonate CO2 paleobarometer by refining input parameters (e.g., soil temperature, soil CO2 production function) have yielded more accurate estimates of ancient atmospheric pCO2. The carbonate CO2 paleobarometer is especially sensitive to input values for the carbon-isotope composition of soil organic matter (OM), which should ideally reflect the δ13C value of OM present when pedogenic carbonates were forming. Published soil pCO2 estimates derived from pedogenic carbonate in Upper Jurassic paleosols are re-evaluated here using OM occluded within pedogenic carbonates rather than average values of fossilized plant material collected throughout the sampled stratigraphic sections. The new soil pCO2 estimates calculated using occluded OM range from 4,600 to 20,000 ppmV and are much lower than the previously published estimates, which were often in excess of 60,000 ppmV. In order to determine which OM source provides more accurate results, estimates of atmospheric pCO2 obtained using plant material and occluded OM from a carbonate-bearing modern soil are compared with measured, pre-industrial atmospheric CO2 levels. In the modern soil profile, plant OM δ13C is highly variable and slightly more negative than the δ13C of occluded OM. The observed ∼ 1‰ offset between the average δ13C values of plant material and occluded OM is much less than the overall range of ∼ 6‰ in plant OM. Estimates of atmospheric pCO2 from the modern soil that are calculated using occluded OM differ by less than 100 ppmV, on average, from estimates based on plant OM. These results suggest that occluded OM produces reasonably accurate pCO2 estimates when used with the carbonate CO2 paleobarometer. Applying a −1‰ correction to the δ13C of occluded OM also produces accurate pCO2 estimates, but the extreme variability in δ13C of plant matter leads to inaccurate pCO2 estimates, even when samples are averaged.