Effects of heavy, moderate, and rare alcohol consumption on fetal development were analyzed in a prospective study of 469 mother-infant pairs. Differential effects of heavy drinking in early and late gestation were evaluated by separate analysis of neonates born to women who reduced consumption before the third trimester. Using chi 2 analysis, multiple regression, and matched sets, statistically significant associations (P less than .01) were observed between sustained heavy drinking and both intrauterine growth retardation and congenital anomalies. These associations were independent of eight other risk factors. No differences were observed between offspring of rare and moderate drinkers. Infants born to women who reduced heavy drinking did not differ in growth from offspring of rare and moderate drinkers but demonstrated a higher frequency of abnormalities. Sustained heavy drinking represents a major risk; reduction in midpregnancy can benefit the newborn. Identification and therapy of heavy drinking are important components of prenatal care.