The association between clinical eating disorders, maternal body weight, shape, and eating concerns, and the birth of low-birth-weight infants (LBW; less than 2500 g) was investigated using a retrospective case-control study. Eighty-eight women delivering LBW infants were interviewed and then divided into two groups--those delivering term, small-for-gestational-age infants (SGA; 37 or more completed weeks, n = 34) and those delivering premature infants (less than 37 completed weeks, n = 54). There were 86 reference women (CTRL) matched for age, parity, and health insurance status, who delivered babies with birth weights greater than 2500 g. In the week postpartum, women delivering term SGA, premature (PREM), and CTRL infants were interviewed using a semistructured interview. One section of this interview included a modified version of the Eating Disorder Examination (EDE), which retrospectively generated, over the previous 12 months, diagnosis of an eating disorder and maternal "normative" weight and shape concerns. In the 3 months before pregnancy, 32% of SGA women, 9% PREM women, and 5% of reference women were diagnosed as having a clinical eating disorder. Women with a past history of an eating disorder had no greater risk of delivering a low-birth-weight infant. Women delivering SGA infants, reported elevated eating disorder psychopathology postdelivery (Eating Disorders Inventory, EDI) and more disturbances in eating behavior before and during pregnancy. Unique predictors for delivery of a LBW term SGA infant were: low maternal prepregnancy body weight, smoking, low maternal weekly weight gain, and elevated EDI (Bulimia subscale). Unique predictors for delivery of a LBW premature infant were: lower maternal occupational status, vomiting in pregnancy, and lower dietary restraint. Women with disordered eating were shown to be at greater risk of delivering term SGA infants. Predictors of term growth retardation are partly determined by maternal behavior.