The evidence for an association between poor nutrition in early life and subsequent obesity is inconclusive. In the present study, we investigated the associations between stunting, wasting and underweight at 2 and 4 years of age, and body composition in adolescence in male subjects studied since birth. The 1982 Pelotas Birth Cohort Study included all children born in maternity hospitals and living in the urban area of the city of Pelotas, Southern Brazil. All males born in 1982 were legally required to enlist in the army between January and April 2000. We were thus able to track 2250 subjects in 2000 (78·9 % of the original cohort). Anthropometric measurements were collected in 1984 and 1986, and body composition was assessed in 2000. In the present analysis, we used as predictors the nutritional indices height-for-age, weight-for-height and weight-for-age presented in six categories. Outcomes included fat, lean and body mass indices and fat:lean mass ratio, derived from anthropometric and bioimpedance measurements. ANOVA and linear regression were used in the analyses to adjust for confounding. All predictors were positively associated with fat and body mass indices. Height-for-age Z score at age 2 or 4 years was not associated with lean mass index, but all other predictors were associated. Fat:lean mass ratio was associated only with weight-for-height Z score. Our results suggest that undernutrition is not a risk factor for overweight and obesity in our population and may partially protect against fatness in adolescence.