In the past, variations in the daily intake of an individual maintaining his weight and activity were considered negligible as it was assumed they arose from measurement errors. It was therefore believed that such variation could be eliminated by averaging over an appropriate period. This, however, is not borne out by the experimental data. The present study reports findings based on a longitudinal diet survey conducted on 43 adolescent schoolboys in a village, over a period of 1 year. A diet survey was carried out for 1 week every 3 months. Intake was estimated by recording the weights of all the food items actually consumed during the day by the students and analysing corresponding food samples in the laboratory. Intake data when analysed for the relative contributions of between- and within-subject variation showed that not only intra-individual variation contributes a substantial portion of the total variation but a significant part of it arises from changes in the mean intake from week to week. The daily intakes of a subsample of 5 students were observed for 49 consecutive d to study the short-term nature of the variation. The results obtained confirm those from the long-term study. Additionally there is an indication that daily intakes of an individual are serially correlated.