Closure of the surface energy balance provides an objective criterion for evaluating eddy-covariance (EC) flux measurements. This study analyses 5 years of EC carbon dioxide, water vapor, and sensible heat flux measurements from three mature boreal forest stands in central Saskatchewan, Canada. The EC sensible and latent heat fluxes, H and λE, underestimated the surface available energy by 11% (aspen), 15% (black spruce), and 14% (jack pine). At all sites, the energy-closure fraction CF responded similarly to the friction velocity u * , atmospheric stability, and time of day. At night, CF increased from ∼0.3 at very low-u * to an asymptotic maximum of ∼0.9 at u * above 0.35ms −1 . During unstable-daytime periods, CF varied linearly from ∼0.7 at low-u * to ∼1.0 at high-u * . The energy imbalance pattern was similar among sites and may be characteristic of the continental, boreal forest.EC measurements of net ecosystem exchange F NEE have no objective, diagnostic parameter that is equivalent to CF. We therefore derived an analogous F NEE “closure fraction” CF NEE by normalizing measured F NEE against estimates from an empirical model that was tuned to the high-u * data. CF and CF NEE responded similarly to u * , atmospheric stability, and time of day. We discuss two implications for EC flux data post-processing. The results uphold the common practice of rejecting EC measurements during low-u * periods. They also lend support to the application of energy-closure adjustments to H, λE, and F NEE .