We determined O2:CO2 exchange ratios of ecosystem fluxes during field campaigns in different forest ecosystems (Harvard Forest/United States, Griffin Forest/United Kingdom, Hainich/Germany). The exchange ratios of net assimilation observed in chamber experiments varied between 0.7 and 1.6, with averages of 1.1 to 1.2. A measurement of soil gas exchange yielded an exchange ratio of 0.94. On the other hand, the observed canopy air O2:CO2 ratios, derived from the concurrent variations of O2 and CO2 abundances in canopy air, were virtually indistinguishable from 1.0 over the full diurnal cycle. Simulations with a simple one-box model imply that the combined processes of assimilation, respiration, and turbulent exchange yield canopy air O2:CO2 ratios that differ from the exchange ratios of the separate fluxes. In particular, the simulated canopy air O2:CO2 ratios (1.01 to 1.12) were clearly lower than the exchange ratios of net turbulent fluxes between the ecosystem and the atmosphere (1.26 to 1.38). The simulated canopy air ratios were also sensitive to changes in the regional O2:CO2 ratio of air above the canopy. Offsets between the various exchange ratios could thus arise if the component ecosystem fluxes have different diurnal cycles and distinct exchange ratios. Our results indicate that measurements of O2 and CO2 abundances in canopy air may not be the appropriate method to determine O2:CO2 exchange ratios of net ecosystem fluxes.