Although models of branching in arterial and bronchial trees often predict a dependence of bifurcation parameters on the scale of the bifurcating vessels, direct verifications of this dependence by comparison with data are uncommon. We compare measurements of bifurcation parameters of airways and arterial trees of different mammals as a function of scale to general features predicted by theoretical models based on minimization of pumping power and network volume. We find that the size dependence is more complex than existing theories based solely on energy and volume minimization, and suggest additional factors that may govern the branching at different scales.The potential factors determining the parameters at bifurcations in the arterial and bronchial trees of mammals have lately received attention. Most notably, these local bifurcation characteristics have been proposed to determine allometric scaling laws for biologically important variables (see [1, 2] and [3]). It has been suggested that arterial and bronchial trees follow a pattern that minimizes a combination of pumping power and the total volume of the tree (see for example [1]-[6]). If the flow is assumed to be Poiseuille (resistance proportional to the inverse fourth power of the vessel radius) on all scales, self-similar branching results, i.e., the branching parameters do not depend on the scale. However, it has been argued [1] that, due to the fact that the blood flow is pulsatile [8,9], arterial trees determined by optimization have branching parameters that depend on the scale (the form of this dependence is discussed below). As a direct test of these theories and their assumptions, we compare in this Letter experimental measurements of the scale dependence of the branching parameters with the predictions of theoretical models. Many studies of bifurcation parameters rely on ordering schemes [10] which do not directly consider the dependence of the bifurcation parameters on the scale, and thus are not appropriate for our particular purpose. We will explicitly study the dependence of the radii of the bifurcated vessels on the radius of the parent vessel, thus allowing a more direct comparison with theoretical models. We show that important aspects of previous observations [11] for the human bronchial tree, not predicted by the theoretical models, are more general than has been previously noticed. In particular, we find qualitatively similar behavior for dog arterial and bronchial trees, and lamb fetal and neonatal arterial trees. Our results suggest that, in addition to minimizing pumping power and network volume, there might be other important factors determining the * Electronic address: juanga@math.umd.edu branching in the arterial and bronchial trees.Arterial and bronchial trees start with a tree root (a parent vessel) that divides into two (daughter) vessels, and these in turn subdivide in a similar fashion, continuing until a certain approximate value of the radius is attained (in the case of arterial trees, the capillary radius). At a partic...