Objective: To determine the association between household food security and infant complementary feeding practices in rural Bangladesh. Design: Prospective, cohort study using structured home interviews during pregnancy and 3 and 9 months after delivery. We used two indicators of household food security at 3-months' follow-up: maternal Food Composition Score (FCS), calculated via the World Food Programme method, and an HHFS index created from an eleven-item food security questionnaire. Infant feeding practices were characterized using WHO definitions. Setting: Two rural sub-districts of Kishoreganj, Bangladesh. Subjects: Mother-child dyads (n 2073) who completed the 9-months' follow-up. Results: Complementary feeding was initiated at age ≤4 months for 7 %, at 5-6 months for 49 % and at ≥7 months for 44 % of infants. Based on 24 h dietary recall, 98 % of infants were still breast-feeding at age 9 months, and 16 % received ≥4 food groups and ≥4 meals (minimally acceptable diet) in addition to breast milk. Mothers' diet was more diverse than infants'. The odds of receiving a minimally acceptable diet for infants living in most food-secure households were three times those for infants living in least food-secure households (adjusted OR = 3·0; 95 % CI 2·1, 4·3). Socio-economic status, maternal age, literacy, parity and infant sex were not associated with infant diet. Conclusions: HHFS and maternal FCS were significant predictors of subsequent infant feeding practices. Nevertheless, even the more food-secure households had poor infant diet. Interventions aimed at improving infant nutritional status need to focus on both complementary food provision and education.