BACKGROUND: To address limited longitudinal nutrition data on children and adolescents, a self-administered food frequency questionnaire was designed for older children and adolescents. Initially, the Youth/Adolescent Questionnaire (YAQ) was developed and demonstrated to be reproducible. This study was conducted to evaluate its validity. METHODS: The form was administered twice to a sample of 261 youths (ages 9 to 18) at an approximate interval of 1 year (1993-1994), and three 24-hr dietary recalls were collected during this period. Pearson correlation coefficients were calculated on nutrient data. RESULTS: Validity was first evaluated by comparing the average of the three 24-hr recalls to the average of the two YAQs. Similar mean nutrients were found by both methods. Correlation coefficients between the mean energy-adjusted nutrients computed by the two methods ranged from 0.21 for sodium to 0.58 for folate. After correction for within-person error, the average correlation coefficient was 0.54, similar to that found among adults. CONCLUSION: A simple self-administered questionnaire completed by older children and adolescents can provide nutritional information about this age group.