Early changes in branching geometry of microvasculature and its associated impact on the perfusion distribution in diseases, especially those in which different branching generations are affected differently, require the ability to analyze intact vascular trees over a wide range of scales. Micro-CT offers an excellent framework to analyze the microvascular branching geometry. Such an analysis requires methods to be developed that can accurately characterize branching properties, such as branch diameter, length, branching angle, and branch interconnectivity of the microvasculature. The purpose of this article is to report the results of a study of two human intramyocardial coronary vascular tree casts in which the accuracy of micro-CT vascular imaging and its analysis are tested against measurements made through an optical microscope (used as the "gold-standard"). Methods related to image segmentation of the vascular lumen, vessel tree centerline extraction, individual branch segment measurement, and compensating for the non-ideal modulation transfer function of micro-CT scanners are presented. The extracted centerline accurately characterized the hierarchical structure of the vascular tree casts in terms of "parent-branch" relationships which allowed each interbranch segments' dimensions to be compared to the optical measurement method. The comparison results show a close to ideal 1:1 relationship for both length and diameter measurements made by the two methods. Combining the results from both specimens, the standard deviation of the difference between measurement methods was 19 microm for the measurement of interbranch segment diameters (ranging from 12 to 769 microm), and 172 microm for the measurement of interbranch segment lengths (ranging from 14 to 3252 microm). These results suggest that our micro-CT image analysis method can be used to characterize a vascular tree's hierarchical structure, and accurately measure interbranch segment lengths and diameters.