The E693Q mutation in the amyloid beta precursor protein (APP) leads to cerebral amyloid angiopathy (CAA), with recurrent cerebral hemorrhagic strokes and dementia. In contrast to Alzheimer disease (AD), the brains of those affected by hereditary cerebral hemorrhage with amyloidosis-Dutch type (HCHWA-D) show few parenchymal amyloid plaques. We found that neuronal overexpression of human E693Q APP in mice (APPDutch mice) caused extensive CAA, smooth muscle cell degeneration, hemorrhages and neuroinflammation. In contrast, overexpression of human wild-type APP (APPwt mice) resulted in predominantly parenchymal amyloidosis, similar to that seen in AD. In APPDutch mice and HCHWA-D human brain, the ratio of the amyloid-beta40 peptide (Abeta40) to Abeta42 was significantly higher than that seen in APPwt mice or AD human brain. Genetically shifting the ratio of AbetaDutch40/AbetaDutch42 toward AbetaDutch42 by crossing APPDutch mice with transgenic mice producing mutated presenilin-1 redistributed the amyloid pathology from the vasculature to the parenchyma. The understanding that different Abeta species can drive amyloid pathology in different cerebral compartments has implications for current anti-amyloid therapeutic strategies. This HCHWA-D mouse model is the first to develop robust CAA in the absence of parenchymal amyloid, highlighting the key role of neuronally produced Abeta to vascular amyloid pathology and emphasizing the differing roles of Abeta40 and Abeta42 in vascular and parenchymal amyloid pathology.