Evidence for atherosclerosis reversal comes from studies in animals wherein atherosclerosis is induced and then allowed to regress, autopsy studies of starved humans, and angiographic studies testing antiatherosclerosis treatment. Animal models and autopsy studies have provided detailed histologic and biochemical descriptions of regression. Cellular and subcellular information exists on what can occur, but because the same lesions are not re-examined, what actually does occur is unknown. Studies of isolated arterial cell systems and intact lesions indicate that atherogenesis involves at least the following: Increased permeability of the endothelium to macromolecules such as low-density lipoprotein; platelet adherence to areas of functional endothelial injury or denudation; the entrance of monocytes or macrophages and lymphocytes into the subintimal space; and the secretion of growth factors by platelets, injured endothelium, and macrophages. These processes can be initiated or enhanced by various vasoactive agents that induce endothelial cell constriction with the opening of endothelial junctions. These processes also can recruit smooth muscle cells from the media to the subintima where they proliferate. Proliferating smooth muscle cells, along with macrophages, can internalize lipids and lipoproteins to form foam cells. Subintimal smooth muscle cells can also synthesize collagen, elastin, glycosaminoglycans, and other connective tissue elements that trap lipoproteins. Peroxidative injury increases the atherogenic potential of both cholesteryl ester-rich (low-density) and triglyceride-rich (very-low-density and intermediate-density) lipoproteins. Steep oxygen gradients within the arterial wall create local conditions for free radical generation, and any increase in residence time of lipoprotein particles can be atherogenic. In summary, there are many areas where treatment may retard or reverse atherogenesis. Angiographic trials that identify and track individual human lesions have shown that reducing known atherogenic risk factors can lessen coronary and femoral atherosclerosis. But they provide no information on events within arterial wall cells or the intracellular matrix. They deal only with lesions that intrude into the vessel lumen and obtain measurements at infrequent intervals. The weight of evidence is that regression is possible, but there is no consensus on the most effective therapy. The challenge for future trials is to select optimal targets for intervention among the known atherogenic processes.