For over a century, naturalists have marvelled over the spectacular variation in plant form and function that characterizes Earth's biomes. Only recently, have the data and tools become available to enable a synoptic and quantitatively rigorous assessment of how plant functional diversity (FD) varies along broad geographic and climatic gradients. The need for such an analysis is more pressing than ever as ecologists attempt to understand how FD mediates ecosystem responses to climate change. Here we combine several plant functional trait databases with a large plant distributional database to examine, for the first time, regional scale variation in plant FD along geographic and climatic gradients. Using null modelling techniques to control for variation in regional species richness we present the first intercontinental map of plant FD and test the hypotheses that regional scale plant FD is: (i) directly linked to species richness (the sampling effect hypothesis); and (ii) lower in less favourable climates (the favourability hypothesis). Our results show that gradients in FD are not simply the result of a sampling effect and that FD is greatly reduced in regions with large annual temperature ranges and low variance in monthly rainfall. Thus, controlling for regional species richness, seasonality of temperature and precipitation have opposite effects on FD. Ultimately the present work provides novel and critical baseline understanding relevant to ecological problems ranging from modelling plant community assembly to predicting ecosystem resilience to climate change.