The local injection of pure inflammatory mediators induces venular leakage. To test the effect of endogenous mediators from dying tissue on vascular leakage, the authors devised an experimental model simulating an infarct, whereby living vessels would be exposed to fragments of organs undergoing aseptic necrosis. Tissues from donor rats were implanted aseptically in the cremasteric sac. Control rats were implanted with materials deemed to be as close as possible to nonirritating: boiled tissues and spheres of Teflon or glass. At different points the rats were injected intravenously with carbon black and killed an hour later. Whole cremaster mounts showed that vascular labeling was strictly venular up to 8 hours, mixed with capillary labeling between 12 and 24 hours, and mainly or exclusively capillary at 48 hours. Histology showed an acute inflammatory infiltrate in the labeled areas. A similar but weaker labeling pattern accompanied by milder inflammation was seen in controls. These results indicate that the vascular leakage in aseptic inflammation is biphasic, first venular, then capillary; and that the capillary phase is induced by the inflammatory reaction itself, possibly through a form of diffuse angiogenesis.