Stunting is a measure of chronic undernutrition, and it affects approximately 160 million children worldwide. Cognitive development of stunted children is compromised, but evidence about the association between height gain in late childhood and adolescent cognitive capacity is scarce. We aimed to determine the association between height gains at different ages, including late childhood, and cognitive capacity at 15-years-of-age. We conducted a prospective cohort study in a rural African setting in Southern Malawi. The study cohort was enrolled between June 1995 and August 1996. It originally comprised mothers of 813 fetuses, and the number of children born live was 767. These children were followed up until the age of 15 years. The anthropometrics were measured at one and 24-months-of-age and 15-years-of-age, and cognitive capacity of participants was assessed at 15-years-of-age with Raven's Coloured Matrices score, mathematic test score, median reaction time (RT) (milliseconds) and RT lapses. The associations between growth and the outcome measures were assessed with linear regression. Raven's Coloured Matrices score was predicted by height gain between 24 months and 15-years-of-age (coefficient 0.85, P = 0.03) and (coefficient 0.69, P = 0.06), but not by earlier growth, when possible confounders were included in the model. The association weakened when school education was further added in the model (coefficient = 0.69, P = 0,060). In conclusion, in rural Malawi, better growth in late childhood is likely to lead to better cognitive capacity in adolescence, partly through more school education. In light of these results, growth promotion should not only be limited to early childhood.