[Truncated abstract] One of the barriers to understanding and preventing proliferative diabetic retinopathy in humans has been the lack of an appropriate animal model. Historically dog, rat and mouse models of diabetic retinopathy have been studied but none of these exhibit the later changes of proliferative diabetic retinopathy. Animals can be rendered diabetic by surgical pancreatectomy or the use of chemicals such as allozan or streptozotocin or by feeding of a high galactose diet. Alternatively, spontaneous rodent models of diabetes have been examined such as the BB rat, KK mouse or NOD mouse. However, in each case the retinal vascular changes observed are those of early nonproliferative diabetic retinopathy comprising at most saccular microaneurysms, increased thickness of the capillary basement membrane, acellular capillaries and pericyte ghosts. … Fluorecein angiography of this transgenic line clearly demonstrates the presence of leaky new vessels, by the appearance of leakage spots scattered throughout the retina from 1 month of age. These mice constitute a valuable model of diabetic retinopathy. Neovascularization in this animal model is induced by VEGF as in human diabetic retinopathy. The source of VEGF in human diabetic retinopathy is the ischemic inner retina. In this transgenic model the source of VEGF are the photoreceptor cells, which are situated just underneath the inner retina. The neovascularization is not dependent on a particular developmental stage and there is no spontaneous regression of new vessels. Thus any results generated in this model are highly relevant to human diabetic retinopathy.