This study was conducted to evaluate the influence of demographic characteristics, hospital practices, maternal psychosocial factors, and knowledge about infant feeding and breast milk on duration of breastfeeding. The mothers of 91 healthy, term infants delivered at a university hospital between June 1998 and December 1998, and first seen in the well-child unit within 10 days of delivery, participated in the study. Forty-nine (54%) infants were exclusively breastfed at 4 months of age. Cox regression analysis showed a negative association between formula supplementation during the hospital stay and duration of exclusive breastfeeding. The median age for starting non-breast milk liquids was 1 month for those who received formula in the hospital and 3 months for those who did not (P = .001). The hospital practices were more predictive of the duration of exclusive breastfeeding in this study group than mothers' knowledge of infant feeding or psychosocial factors. Thus, hospital practices should be reevaluated.