We aimed to comprehensively examine the association of breast-feeding, types and initial timing of complementary foods with adolescent cognitive development in low- and middle-income countries. We conducted a prospective cohort study of 745 adolescents aged 10–12 years who were born to women who participated in a randomised trial of prenatal micronutrient supplementation in rural Western China. An infant feeding index was constructed based on the current WHO recommendations. Full-scale intelligence quotient (FSIQ) was assessed and derived by the fourth edition of the Wechsler Intelligence Scale for Children. The duration of exclusive or any breast-feeding was not significantly associated with adolescent cognitive development. Participants who regularly consumed Fe-rich or Fe-fortified foods during 6–23 months of age had higher FSIQ than those who did not (adjusted mean differences 4·25; 95 % CI 1·99, 6·51). For cows’/goats’ milk and high protein-based food, the highest FSIQ was found in participants who initially consumed at 10–12 and 7–9 months, respectively. A strong dose–response relationship of the composite infant feeding index was also identified, with participants in the highest tertile of overall feeding quality having 3·03 (95 % CI 1·37, 4·70) points higher FSIQ than those in the lowest tertile. These findings suggest that appropriate infant feeding practices (breast-feeding plus timely introduction of appropriate complementary foods) were associated with significantly improved early adolescent cognitive development scores in rural China. In addition, improvement in Fe-rich or Fe-fortified foods complementary feeding may produce better adolescent cognitive development outcomes.