In this Letter, we demonstrate high resolution, three-dimensional optical imaging of in vivo blood vessel networks using speckle variance optical coherence tomography, and the quantification of these images through the development of biologically relevant metrics using image processing and segmentation techniques. Extracted three-dimensional metrics include vascular density, vessel tortuosity, vascular network fractal dimension, and tissue vascularity. We demonstrate the ability of this quantitative imaging approach to characterize normal and tumor vascular networks in a preclinical animal model and the potential for quantitative, longitudinal vascular treatment response monitoring.