Self-reported and measured height and weight were obtained from a representative sample of 1,598 persons in Auckland, New Zealand during 1982. The accuracy of the self-reported data and its effect on the misclassification of relative weight, as measured by Quetelet index, were examined. The finding that for most participants (75%), self-reported measures were no more than 3.5 cm from their measured height and 2.4 kg from their measured weight indicates that self-reports have a high degree of accuracy. However, the participants consistently overestimated their height and underestimated their weight, resulting in an underestimation of relative weight. This would have little effect on analyses using the self-reported relative weight measures as a continuous covariate, but misclassification would occur when using relative weight as a categorical variable. The sensitivities and specificities associated with categorized self-reported relative risks that have been calculated from relative weight derived from self-reported height and weight.