BACKGROUND The classical risk approach to predicting who benefits from an intervention is unsound because it relies on the theoretical assumption that those at risk will necessarily benefit. A better approach to systematically test who benefits from nutrition supplementation is proposed using interactive models. METHODS Differential effects of nutrition supplementation during early childhood on stature at adolescence were studied in 245 males and 215 females to identify determinants of long-term benefit from food supplementation. Factors studied included family socioeconomic status (SES) and children's home diet and diarrhoea during the first 3 years of life. To determine whether a factor conferred benefit, the statistical significance of the interaction between this factor and the intervention was tested. Data from the INCAP supplementation trial in Guatemala and from the follow-up of the same subjects at adolescence were used. RESULTS Ordinary least squares (OLS) showed that high rates of diarrhoea in males and poor SES in females were significant determinants of benefit from supplementation at adolescence, and that the effects were mediated by length at 3 years old. Results of two-stage least squares (2SLS) analysis showed that length at 36 months, maturation and maternal height were significant determinants of height at adolescence but SES was not. CONCLUSIONS Nutrition supplementation in early childhood has long-lasting effects on body size and the larger benefits acquired by some groups of children remain throughout early adulthood. The relevance of these findings for screening and targeting of nutritional interventions is discussed.