Estimating leaf temperature distributions (LTDs) in canopies is crucial in forest ecology. Leaf temperature affects the exchange of heat, water, and gases, and it alters the performance of leaf-dwelling species such as arthropods, including pests and invaders. LTDs provide spatial variation that may allow arthropods to thermoregulate in the face of long-term changes in mean temperature or incidence of extreme temperatures. Yet, recording LTDs for entire canopies remains challenging. Here, we use an energy-exchange model (RATP) to examine the relative roles of climatic, structural, and physiological factors in influencing three-dimensional LTDs in tree canopies. A Morris sensitivity analysis of 13 parameters showed, not surprisingly, that climatic factors had the greatest overall effect on LTDs. In addition, however, structural parameters had greater effects on LTDs than did leaf physiological parameters. Our results suggest that it is possible to infer forest canopy LTDs from the LTDs measured or simulated just at the surface of the canopy cover over a reasonable range of parameter values. This conclusion suggests that remote sensing data can be used to estimate 3D patterns of temperature variation from 2D images of vegetation surface temperatures. Synthesis and applications. Estimating the effects of LTDs on natural plant-insect communities will require extending canopy models beyond their current focus on individual species or crops. These models, however, contain many parameters, and applying the models to new species or to mixed natural canopies depends on identifying the parameters that matter most. Our results suggest that canopy structural parameters are more important determinants of LTDs than are the physiological parameters that tend to receive the most empirical attention.