White matter hyperintensities (WMH) are commonly seen on neuroimaging scans, but their underlying histopathologic substrate is unclear. The aim of this work was to establish the pathologic correlates of WMH in unselected elderly cases using two study designs. To avoid potential bias from comparisons of different anatomic regions, study 1 compared, region-by-region, the severity of WMH determined in vivo with measures of each of the major white matter (WM) components. Study 2 compared the histopathology of WMH with normal WM. Study 1: The periventricular and deep WM regions of three lobes in 23 brains with in vivo MRI scans were investigated using histologic and immunohistochemical stains. The severity of each pathologic measure was correlated with WMH severity determined using the Scheltens scale. Study 2: Lesioned and nonlesioned areas identified by postmortem MRI in the frontal WM of 20 brains were examined histologically and immunohistochemically. No single pathologic variable correlated with the severity of WMH; however, a multiple stepwise regression analysis revealed that vascular integrity predicted total Scheltens score (beta = -0.53, p = 0.01). Comparison of lesioned and nonlesioned areas demonstrated that vascular integrity was reduced in WMH [t(18) = 3.79, p = 0.001]. Blood-brain barrier integrity was also found to be reduced in WMH [t(5) = -5.31, p = 0.003]. White matter hyperintensities (WMH) involve a loss of vascular integrity, confirming the vascular origin of these lesions. This damage to the vasculature may in turn impair blood-brain barrier integrity and be one mechanism by which WMH evolve.