In many forms of experimental kidney diseases, renal VEGF is low, and administering VEGF can be shown to be protective. A paradox occurs in diabetes, in which renal VEGF levels are high and a deleterious effect of VEGF on kidney disease has been shown. We have hypothesized that endothelial dysfunction induced by hyperglycemia or other factors may underlie the pathogenic mechanisms of a high VEGF state. VEGF normally stimulates endothelial nitric oxide (NO) release and acts in concert with elevated NO levels as a trophic factor for vascular endothelium. The increased NO derived from the endothelial cell acts as an inhibitory factor that prevents excess endothelial cell proliferation, vascular smooth muscle cell proliferation, and macrophage infiltration. In the setting where NO bioavailability is reduced in diabetes, high levels of VEGF lead to excessive endothelial cell proliferation, stimulation of macrophage chemotaxis, and vascular smooth muscle cell activation. Consistent with this hypothesis is our recent observation that diabetes induced in endothelial NO-deficient mice results in clinical and histological features identical to human diabetic nephropathy. The discovery of the key role for impaired endothelial NO bioavailability in the stimulation of VEGF and VEGF-dependent disease may provide key insights into not only the pathogenesis of diabetic nephropathy but also the utility and hazard of administering VEGF as a treatment for kidney disease.