The relationship of maternal use of marijuana and cocaine during pregnancy to measures of neonatal body proportionality and body composition was assessed in a multiethnic sample of 1082 newborn infants. Maternal use of marijuana and cocaine during pregnancy was ascertained by self-report and by an enzyme-multiplied immunoassay technique for screening of urine samples obtained prenatally and again post partum. After each substance was analytically controlled for use of the other and for other potentially confounding variables, detection of marijuana metabolites in maternal urine was associated (p less than 0.05) with depressed mean arm muscle circumference and nonfat area of the arm but not with any measure of neonatal fatness. In contrast, detection of cocaine in maternal urine was associated (p less than 0.05) with decrements of subscapular fat folds and of the fat and nonfat areas of the arm. Although both substances were associated with depressed birth weight, there was no decrement of neonatal ponderal index or of the arm circumference/head circumference ratio in association with exposure to either substance. We conclude that both marijuana exposure and cocaine exposure during pregnancy are associated with symmetric intrauterine growth retardation, but that deficits are in differing compartments of intrauterine growth. These findings suggest that marijuana may retard fetal growth through maternal-fetal hypoxia, whereas cocaine may alter nutrient transfer to the fetus and fetal metabolism.