The joint associations of maternal cigarette smoking and social class on perinatal outcome were studied in the 1970 British birth cohort (British Births). Whereas smoking was much more frequent among women in social classes III, IV, and V, there was little difference in the birthweight decrement associated with smoking across class. Perinatal mortality, however, was increased only among smokers in the manual social classes. Thus whereas the offspring of more privileged smokers were not protected from intrauterine growth retardation, they did not suffer from increased perinatal mortality. Observations of other populations suggest a possible nutritional mediation of this protective effect.