Precise determination of canopy conductance (g(s)) is needed to quantify the water loss and CO2 exchange from forest canopies and their response to changing environmental conditions. In this study, we combined measurements of the leaf-to-air vapour pressure difference (D-L) derived from canopy surface temperature, and tree transpiration to calculate canopy g(s) in a pine forest at the ICOS site FR-Bil. The period covered was characterized by two consecutive droughts. The inversion of the generalised water transport equation (GT), along with its isothermal simplification (GT'), were used to calculate canopy g(s) and compared with gs determined from the inversion of the Penman-Monteith equation (PMT). A thermal infrared camera continuously monitored the canopy temperature and allowed to assess the time course of DL with half-hourly resolution. On average a 0.3 degrees C temperature difference was found between the canopy and surrounding air, with values ranging from -2 degrees C to +5 degrees C depending on the time of day, soil moisture and humidity. The three methods used to calculate gs, GT, GT' and PMT were in agreement under wet soil and low atmospheric demand, but differences up to 40% were found under water stress conditions when the canopy to air temperature differences led to substantial discrepancies between D-L and air saturation vapour pressure deficit at 8.2 m in the crown (D-8). Under such conditions the GT' method overestimated g(s) whereas the PMT method was closer from values estimated with the GT method. The contrasted behaviour of the atmospheric exchanges between the tree canopy and the overall ecosystem limits the use of the above canopy flux measurements alone to quantify the response of surface conductance to environmental drivers. Our results also advocate the use of in situ surface temperature measurements to better understand the response of plant canopies to environmental conditions.< Réduire