Social selection, or the idea that an individual's health can influence their social mobility and, hence, their position in the social hierarchy, has been suggested as an important element in the process which produces social class differences in health. The present paper examines this idea by drawing together evidence from a range of published research. Direct selection according to health is judged to have little effect on class gradients. The logically distinct idea of indirect selection can be seen most usefully as referring to the accumulation of advantage or disadvantage during life.