Ecosystem water-use efficiency (eWUE; the ratio of net ecosystem productivity to evapotranspiration rate) is a complex landscape-scale parameter controlled by both physical and biological processes occurring in soil and plants. Leaf WUE (lWUE; the ratio of leaf CO2 assimilation rate to transpiration rate) is controlled at short time scales principally by leaf stomatal dynamics and this control varies among plant species. Little is known about how leaf-scale variation in lWUE influences landscape-scale variation in eWUE. We analyzed approximately seven thousand 30-min averaged eddy covariance observations distributed across 9 years in order to assess eWUE in two neighboring forest communities. Mean eWUE was 19% lower for the community in which Engelmann spruce and subalpine fir were dominant, compared to the community in which lodgepole pine was dominant. Of that 19% difference, 8% was attributed to residual bias in the analysis that favored periods with slightly drier winds for the spruce-fir community. In an effort to explain the remaining 11% difference, we assessed patterns in lWUE using C isotope ratios. When we focused on bulk tissue from older needles we detected significant differences in lWUE among tree species and between upper and lower canopy needles. However, when these differences were scaled to reflect vertical and horizontal leaf area distributions within the two communities, they provided no power to explain differences in eWUE that we observed in the eddy covariance data. When we focused only on bulk needle tissue of current-year needles for 3 of the 9 years, we also observed differences in lWUE among species and in needles from upper and lower parts of the canopy. When these differences in lWUE were scaled to reflect leaf area distributions within the two communities, we were able to explain 6.3% of the differences in eWUE in 1 year (2006), but there was no power to explain differences in the other 2 years (2003 and 2007). When we examined sugars extracted from needles at 3 different times during the growing season of 2007, we could explain 3.8–6.0% of the differences in eWUE between the two communities, but the difference in eWUE obtained from the eddy covariance record, and averaged over the growing season for this single year, was 32%. Thus, overall, after accounting for species effects on lWUE, we could explain little of the difference in eWUE between the two forest communities observed in the eddy covariance record. It is likely that water and C fluxes from soil, understory plants, and non-needle tissues, account for most of the differences observed in the eddy covariance data. For those cases where we could explain some of the difference in eWUE on the basis of species effects, we partitioned the scaled patterns in lWUE into two components: a component that is independent of canopy leaf area distribution, and therefore only dependent on species-specific differences in needle physiology; and a component that is independent of species differences in needle physiology, and only dependent on species-specific influences on canopy leaf area distribution. Only the component that is dependent on species influences on canopy leaf area distribution, and independent of inherent species differences in needle physiology, had potential to explain differences in eWUE between the two communities. Thus, when tree species effects are important, canopy structure, rather than species-specific needle physiology, has more potential to explain patterns in eWUE.