We aim to identify the importance of vapour pressure deficit (VPD), soil water content (SWC) and photosynthetic photon flux density (PPFD) as drivers of tree canopy conductance, which is a key source of uncertainty for modelling vegetation responses under climate change. We use sap flow time series of 1858 trees in 122 sites from the SAPFLUXNET global database to obtain whole-tree canopy conductance (G). The driver importance, defined as the percentage of variance (R2) of G explained by the three main hydrometeorological drivers (VPD, SWC and PPFD), was evaluated using linear mixed models. For each driver we assess differences in their importance among biomes, and use multiple linear regression to explain how driver importance varies with climate, soil and vegetation structure. We found that in most areas tree canopy conductance is better explained by VPD than by SWC or PPFD, although results for drylands were inconclusive. Climate, soil and vegetation structure were common controls for the importance of all three hydrometeorological drivers, with wetter climates, fine-textured soils and tall vegetation being associated with stronger influence of the driver on G. Differences across sites in the importance of the hydrometeorological drivers of tree canopy conductance may affect predictions of ecosystem dynamics under future climates, and should be accounted for explicitly in models.