We collected self-reported weights and heights of 806 sixth-graders in Pennsylvania and compared these values with measures of the children obtained by trained personnel who used standard clinical scales and rigid stadiometers. Approximately 10% of the children either were unable to specify a value for weight, height, or both or reported values of a magnitude indicating that they could not quantitatively conceptualize these measurements. Of the remaining 90%, both boys and girls tended to underreport their weight. Underreporting weight was more pronounced among girls than among boys; for both sexes, the degree of underreporting was much greater among taller, heavier children than among shorter, lighter children. Deviation between self-reported and measured height was substantial for many individuals, but there was no consistent trend toward underreporting or overreporting. These findings indicate that children's self-reported weight data may have a systematic bias, particularly if a large proportion of the sample is over-weight; height data may have considerable random error. We, therefore, conclude that self-reported weights and heights of children may be subject to errors that could confound results if used in clinical or research efforts.