Accurate estimates of soil CO2 efflux are important in the current discussion on the carbon balance of forests, and can be used to validate models and remotely sensed data. Due to the typically large spatial variability, large sample numbers are required to estimate mean forest soil CO2 efflux with reasonable confidence intervals. Most infrared gas analysis (IRGA) systems are not well suited to simultaneously produce daily means and handle this spatial variability problem. The soda lime technique gives daily means and allows the required large sample numbers, but is less accurate than the IRGA systems. Using an elaborate cross-calibration, we tried to combine the accuracy of an IRGA method with the spatial integration potential of the soda lime technique. This paper reports on the calibration technique used to improve the accuracy of the soda lime technique and confirms the spatial variability in soil CO2 efflux in a heterogeneous forest.