Background: Cross-sectional studies have shown associations between stunting and overweight; however, there are few prospective studies of stunted children. Objectives: To determine whether stunting before age 2 years is associated with overweight and central adiposity at 17-18 years and whether growth in height among stunted children predicts body mass index (BMI) in late adolescence. Design: Prospective cohort study. Participants: One-hundred and three participants stunted by age 2 years and 64 non-stunted participants (78% of participants enrolled in childhood). Participants were measured in early childhood and at ages 7, 11 and 17 years. Results: Stunted subjects remained shorter and had lower BMIs, smaller skinfolds and circumferences than non-stunted subjects. Overweight (BMI X25 m 2 ) was not significantly different among stunted and non-stunted male subjects (5.2 and 12.5%) but non-stunted female subjects were more likely to be overweight than those who experienced early childhood stunting (11.1 and 34.4%, P ¼ 0.013). Centralization of fat (waist to hip ratio (WHR), subscapular/triceps skinfold ratio (SSF/TSF)) did not differ between stunted and non-stunted groups (mean WHR 0.77 and mean SSF/TSF 1.18 in both groups). Stunted subjects with greater increases in height-for-age for the intervals 3-7 and 7-11 years had higher BMI at age 17 years (P ¼ 0.04 and P ¼ 0.001, respectively). Conclusion: Participants stunted by age 2 years were less likely to be overweight than those who were never stunted. This suggests that cross-sectional studies of the association between stunting and overweight may be misleading. Among stunted children, greater linear growth during mid-to late childhood was associated with greater BMI at age 17 years.