1. The study used stable carbon and oxygen isotopes from paleosol carbonates in loess deposits in eastern Washington state to reconstruct past atmospheric CO2 levels. 2. Two methods were used to calculate CO2 levels, either assuming a CO2 value to determine soil respiration rates or assuming soil respiration rates to determine CO2 levels, since soil respiration is difficult to constrain. 3. Results suggest that when using ice core CO2 records to constrain soil respiration rates, the rates for the last glacial maximum were within expected low ranges, reflecting cold and dry conditions, while holocene rates were higher as expected. However, freely varying soil respiration led to unrealistic high CO2 calculations.