Perceptual constancy has played a significant role in philosophy of perception. It figures in debates about direct realism, color ontology, and the minimal conditions for perceptual representation. Despite this, there is no general consensus about what constancyis. I argue that an adequate account of constancy must distinguish it from three distinct phenomena:meresensory stability through proximal change, perceptualcategorizationof a distal dimension, and stability throughirrelevantproximal change. Standard characterizations of constancy fall short in one or more of these respects. I develop an account (...) of constancy that overcomes these problems. The account has two parts: an analysis of constancy mechanisms, and an analysis of the conditions under which a constancy capacity is exercised. I then employ this account to evaluate whether constancy is a necessary condition for perceptual representation, as some have conjectured. I argue that explanatory practice in perceptual psychology fails to support this view. Rather, it fits better with the weaker principle that representation requires specific tracking of a distal dimension. (shrink)