ABSTRACTStemflow measurements were performed on seven Quercus serrata and canopy structure metrics were obtained to identify factors that affect stemflow. Among the seven stems, two exhibited a large stem lean (> 20°). Although a linear relationship between rainfall and stemflow was shown in all trees, stemflow volume on the highly leaning stems (Nos. 6 and 7) was clearly smaller than those on the others. To extract canopy structure metrics effective on stemflow, GLM analysis was conducted. The objective variable was the slope of the regression line a between rainfall and stemflow. Canopy structure metrics as explanatory variables were the diameter at breast height DBH, tree height H, canopy projection area CPA, stem lean parameters L and Ld, and stem length (SL). L is stem lean from the base to the top and Ld is stem lean at breast height. GLMs were constructed using all combinations of explanatory variables. GLMs that included at least one explanatory variable for which a statistically significant had been shown were selected. First, GLMs constructed using DBH, L, and Ld were selected. The relationship between DBH and a was positive, but the R2 value using all data (R2 = 0.39) was smaller than upon excluding stem Nos. 6 and 7 (R2 = 0.85). This indicated the need to consider DBH and stem lean simultaneously. Second, a GLM constructed only using H was selected. R2 values of DBH with H and SL were 0.57 and 0.72, respectively. According to the definition of H, the more the stem leans, the lower H is measured, at the same value of SL. Thus, H was a parameter that includes the effect of stem lean. Our results indicated that it is necessary to consider stem lean when evaluating the process of canopy interception in deciduous broadleaf forests.