A stemflow (SF) estimation model, based on the correlation between coefficients (a andb) of the regression equation (SF=aP g −b, whereP g is gross rainfall) obtained from sample tree and their diameter at breast height (DBH), is presented. Estimation by the model forSF from two deciduous broad-leaved forests shows favorable agreement with the measurement. Once the equations for regression coefficients are determined, the model has the potential to easily estimateSF from an entire stand through sampling of some representative trees. However, the model is site-specific. In order to mitigate the weakness of this model, the combination of previous results provided generalized equations for the coefficients. The generalized equation adequately estimatedSF of some forests while its estimation was biased for other forests, which indicates thatSF variation between forests is closely related to structural differences. Actually, theSF frequency distribution of two forest stands was significantly different. So, influence of structural factors onSF was also studied using funneling ratio (FR) that integrates the efficiency of trunk and canopy characteristics toSF. Comparison ofFR showed thatSF was significantly correlated to crown-projection area (CPA) and basal area at breast height (BA). Parameters (β 1 andβ 2) within the model were significantly correlated to respective stand density and leaf area index (LAI).