We determined δ13C values associated with canopy gross and net CO2 fluxes from four U.S. sites sampled between 2001 and 2002. Annual mean, flux-weighted δ13C values of net ecosystem CO2 exchange (NEE) were estimated for four contrasting ecosystems (three forests and one grassland) by analyzing daytime flask measurements collected at the top of canopies. Combining δ13C values associated with respiratory and net (respiration minus photosynthesis) fluxes, we demonstrate a method for estimating whole-canopy photosynthetic discrimination against 13C (ΔA) in terrestrial ecosystems directly from field measurements. This experimental approach offers the possibility of examining interannual variability in ΔA from ecosystem δ13C measurements. Our estimated δ13C values associated with photosynthetic fluxes are in agreement with those measured from foliar organic matter for C3 forests, and are within the range bounded by C3 and C4 grasses in a tallgrass prairie. The δ13C associated with NEE fluxes at our C3 forest sites ranges between -27.1 ¿ 2.7 and -28.3 ¿ 2.5?, and is -22.6 ¿ 4.0? at the prairie site. These estimates differ from a previous study, particularly for C3 ecosystems at comparable latitudes. Sensitivity analyses indicate that our estimates of δ13C values of net CO2 fluxes are robust with respect to measurement errors, but can vary depending on the selection of background atmospheric values. Other factors (e.g., drought and sampling footprint) that might have influenced our δ13C measurements and calculations of ΔA are also discussed. Our measurement-based analyses are particularly useful when both latitudinal and longitudinal variations in ΔA are to be considered in the global inversion modeling studies.