One hundred sixteen patients who had had kwashiorkor between the ages of 5 months and 4 4/12 years and 89 controls were the subjects of a 15-year follow-up study of their growth and development. We report the findings of a longitudinal analysis of 53 (30 females) of the ex-patients and 30 (15 females) of the controls, selected because at the end of 15 years of study they conformed to the criteria of (1) adult secondary sexual characteristics, (2) height velocity <2 cm/yr, and (3) completeness of data. Nonlinear curve-fitting techniques were applied to the height and weight data for these subjects using the Preece-Baines model 1 growth function to determine the timing and magnitude of adolescent growth spurts and the associated biologic changes. Both expatients and controls grew below the 25th percentile of British longitudinal standards, but the male ex-patients were heavier, and perhaps talier, than the controls for most of their childhood and adolescence. Velocity curves indicated that the ex-patients had higher pre-adolescent peak increments than the controls and a generally longer growth spurt of reduced magnitude. Two possible explanations are discussed: (1) Garrow and Pike's theory that children with kwashiorkor have a genetic potential for greater physical growth, and (2) a socioeconomic crisis occurring within a family affects the youngest child, who subsequently requires a longer time to recover than do siblings within an improving socioeconomic situation. © 1986 The C. V. Mosby Company.