It has been proposed that the stomatal response to humidity relies on sensing of the transpiration rate itself rather than relative humidity or the saturation deficit per se. We used independent measurements of stomatal conductance (gs), transpiration (E), and leaf-to-air vapour pressure difference (V) in a hybrid poplar canopy to evaluate relationships between gs and E and between gs and V. Relationships between E, V and total vapour phase conductance or crown conductance (gc) were also assessed. Conductance measurements were made on exposed and partially shaded branches over a wide range of incident solar radiation. In exposed branches, gs appeared to decline linearly with increasing E and increasing V at both high and low irradiance. However, in a partially shaded branch, a bimodal relationship between gs and E was observed in which gs continued to decrease after E had reached a maximum value and begun to decrease. The relationship between gs and V for this branch was linear. Plots of gc against E always yielded bimodal or somewhat variable relationships, whereas plots of gc against V were invariably linear. It was not possible to derive a unique relationship between conductance and E or V because prevailing radiation partially determined the operating range for conductance. Normalization of data by radiation served to linearize responses observed within the same day or type of day, but even after normalization, data collected on partly cloudy days could not be used to predict stomatal behaviour on clear days and vice versa. An additional unidentified factor was thus also involved in determining operating ranges of conductance on days with different overall radiation regimes. We suggest that the simplest mechanism to account for the observed humidity responses is stomatal sensing of the epidermal or cuticular transpiration rate rather than the bulk leaf or stomatal transpiration rate.