To determine the effects of a child obesity prevention intervention, beginning in pregnancy, on infant feeding practices in low-income Hispanic families. The Starting Early randomized controlled trial enrolled pregnant women at a third trimester visit. Women (n=533) were randomly allocated to a standard care control group or an intervention group participating in prenatal and postpartum individual nutrition/breastfeeding counseling and subsequent nutrition and parenting support groups coordinated with well-child visits. Outcome measures included infant feeding practices and maternal infant feeding knowledge at infant age 3months, using questions adapted from the Infant Feeding Practices Study II and an infant 24-hour diet recall. A total of 456 families completed 3-month assessments. The intervention group had higher prevalence of exclusive breastfeeding on the 24-hour diet recall (42.7% vs 33.0%, P=.04) compared with controls. The intervention group reported a higher percentage of breastfeeding vs formula feeding per day (mean [SD] 67.7 [39.3] vs 59.7 [39.7], P=.03) and was less likely to introduce complementary foods and liquids compared with controls (6.3% vs 16.7%, P=.001). The intervention group had higher maternal infant feeding knowledge scores (Cohen d, 0.29, 95% CI .10-.48). The effect of Starting Early on breastfeeding was mediated by maternal infant feeding knowledge (Sobel test 2.86, P=.004). Starting Early led to increased exclusive breastfeeding and reduced complementary foods and liquids in 3-month-old infants. Findings document a feasible and effective infrastructure for promoting breastfeeding in families at high risk for obesity in the context of a comprehensive obesity prevention intervention. ClinicalTrials.gov: NCT01541761.