During the summers (July and August) of 2002-2005, we measured interannual variation in maximum carboxylation rate (V(cmax)) within a Fagus crenata Blume crown in relation to climate variables such as air temperature, daytime vapor pressure deficit (VPD) and daily photosynthetic photon flux, leaf nitrogen per unit area (N(a)) and leaf mass per unit area (LMA). Climatic conditions in the summers of 2002-2004 differed markedly, with warm and dry atmospheric conditions in 2002, cool, humid and cloudy conditions in 2003, and warm clear conditions in 2004. Conditions in summer 2005 were intermediate between those of summers 2002 and 2003, and similar to recent (8-year) means. In July, marked interannual variation in V(cmax) was mainly observed in leaves in the high-light environment (relative photon flux > 50%) within the crown. At the crown top, V(cmax) was about twofold higher in 2002 than in 2003, and V(cmax) values in 2004 and 2005 were intermediate between those in 2002 and 2003. In August, although interannual variation in V(cmax) among the years 2003, 2004 and 2005 was less, marked variation between 2002 and the other study years was evident. Multiple regression analysis of V(cmax) against the climate variables revealed that VPD of the previous 10-30 days had a significant influence on variability in V(cmax). Neither N(a), LMA nor leaf CO(2) conductance from the stomata to the carboxylation site explained the variability in V(cmax). Our results indicate that the long-term climatic response of V(cmax) should be considered when estimating forest carbon gain across the year.