A dynamic, closed-chamber infrared gas analysis (IRGA) system (DC-1: CIRAS-1, PP-Systems, Hitchin, U.K.) was compared with three other systems for measuring soil CO(2) efflux: the soda lime technique (SL), the eddy correlation technique (EC), and another dynamic, closed-chamber IRGA system (DC-2: LI-6250, Li-Cor, Inc., Lincoln, NE). Among the four systems, the DC-1 systematically gave the highest flux rates. Relative to DC-1, SL, EC and DC-2 underestimated fluxes by 10, 36 and 46%, respectively. These large and systematic differences highlight uncertainties in comparing fluxes from different sites obtained with different techniques. Although the three chamber methods gave different results, the results were well correlated. The SL technique underestimated soil CO(2) fluxes compared with the DC-1 system, but both methods agreed well when the SL data were corrected for the underestimation at higher fluxes, indicating that inter-site comparisons are possible if techniques are properly crosscalibrated. The EC was the only system that was not well correlated with DC-1. Under low light conditions, EC values were similar to DC-1 estimates, but under high light conditions the EC system seriously underestimated soil fluxes. This was probably because of interference by the photosynthetic activity of a moss layer. Although below-canopy EC fluxes are not necessarily well suited for measuring soil CO(2) efflux in natural forest ecosystems, they provide valuable information about understory gas exchange when used in tandem with soil chambers.