High midday temperatures can depress net photosynthesis. We investigated possible mechanisms underlying this phenomenon in leaves of Eperua grandiflora (Aubl.) Benth. saplings. This tropical tree establishes in small gaps in the rainforest canopy where direct sunlight can raise midday temperatures markedly. We simulated this microclimate in a growth chamber by varying air temperature between 28 and 38 degrees C at constant vapor pressure. A decrease in stomatal conductance in response to an increase in leaf-to-air vapor pressure difference (deltaW) caused by an increase in leaf temperature (Tleaf) was the principal reason for the decrease in net photosynthesis between 28 and 33 degrees C. Net photosynthesis decreased further between 33 and 38 degrees C. Direct effects on mesophyll functioning and indirect effects through deltaW were of similar magnitude in this temperature range. Mitochondrial respiration during photosynthesis was insensitive to Tleaf over the investigated temperature range; it thus did not contribute to midday depression of net photosynthesis. Internal conductance for CO2 diffusion in the leaf, estimated by combined gas exchange and chlorophyll fluorescence measurements, decreased slightly with increasing Tleaf. However, the decrease in photosynthetic rate with increasing Tleaf was larger and thus the difference in CO2 partial pressure between the substomatal cavity and chloroplast was smaller, leading to the conclusion that this factor was not causally involved in midday depression. Carboxylation capacity inferred from the CO2 response of photosynthesis increased between 28 and 33 degrees C, but remained unchanged between 33 and 38 degrees C. Increased oxygenation of ribulose-1,5-bisphosphate relative to its carboxylation and the concomitant increase in photorespiration with increasing Tleaf were thus not compensated by an increase in carboxylation capacity over the higher temperature range. This was the principal reason for the negative effect of high midday temperatures on mesophyll functioning.