The dietary intakes of stunted and non-stunted children were measured on enrollment to a longitudinal study of growth and development. Children aged 9-24 months were recruited by house to house survey of several poor areas of Kingston. All children with height for age less than -2 SD of the NCHS standards and weight for height below the standard median were enrolled. Alternate stunted children were matched for age and sex with the non-stunted child (height for age greater than -1 SD) living nearest. Dietary information was obtained by two 24-h recalls. Mean trainer-interviewer reliability was greater than 90 per cent throughout the study. Correlation between energy intakes on the 2 days was 0.68 (P less than 0.001). Stunted children had a significantly less varied diet, fewer dairy products and fruit than non-stunted children. Protein intakes met requirements. Energy intakes were similar in both groups and approximately 200 kcal below the recommended intake. Energy and protein intakes per kg were significantly higher in the stunted children than in non-stunted children (energy P less than 0.001, protein P less than 0.005). Greater morbidity in the stunted children could account for some, but not all, of this difference. Children with lower weight for height had lower intakes (energy P less than 0.05, protein P less than 0.01). Intakes were greater in children who lived in better housing, with more household possessions and whose mothers had more skilled occupations. Good reliability and the above associations indicate that the data are valid.