One-dimensional multilayer biosphere-atmosphere models (e.g., CANVEG) describe ecosystem carbon dioxide (CO2) and water vapor (H2O) fluxes well when cold temperatures or the hydrologic state of the ecosystem do not induce stomatal closure. To investigate the CANVEG model framework under such conditions, CO2, H2O, and sensible heat fluxes were measured with eddy-covariance methods together with xylem sap flux and leaf-level gas exchange in a 16-year-old (in 1999) southeastern loblolly pine forest. Leaf-level gas exchange measurements, collected over a 3-year period, provided all the necessary biochemical and physiological parameters for the CANVEG model. Using temperature-induced reductions of the biochemical kinetic rate constants, the CANVEG approach closely captures the diurnal patterns of the CO2 and H2O fluxes for two different formulations of the maximum Rubisco catalytic capacity (Vc max)-temperature function, suggesting that the CANVEG approach is not sensitive to Vc max variations for low temperatures. A soil moisture correction (wr) to the Ball-Berry leaf-conductance approach was also proposed and tested. The wr magnitude is consistent with values predicted by a root-xylem hydraulic approach and with leaf-level measurements. The wr correction significantly improves the model's ability to capture diurnal patterns of H2O fluxes for drought conditions. The modeled bulk canopy conductance (Gm) for pine foliage estimated from the CANVEG-modeled multilevel resistance values agreed well with canopy conductance (Gc) independently estimated from pine sap flux measurements. Detailed sensitivity analysis suggests that the leaf-level physiological parameters used in CANVEG are not static. The dynamic property of the conductance parameter, inferred from such sensitivity analysis, was further supported using 3 years of porometry measurements. The CANVEG model also reproduced basic biochemical processes as demonstrated by the agreement between modeled and leaf-level measured Ci/Ca, where Ci and Ca are the intercellular and atmospheric CO2 concentration, respectively. The model estimated that vapor pressure deficit does not vary significantly within the canopy but that Ci/Ca varied by more than 15%. The broader implication of this variation is that big-leaf approaches that compress physiological and biochemical parameters into bulk canopy stomatal properties may be suitable for estimating water vapor flux but biased for CO2 ecosystem fluxes. ¿ 2000 American Geophysical Union