Self-reported height and weight were compared with directly-measured controlled height and weight in 158 women with a median age of 34 years (16-62 years) and in 594 men with a median age of 37 (16-66 years). A model of two one-dimensional linear regressions with two independent variables was used in order to describe the combined influence of both controlled height and controlled weight on self-reported height and on self-reported weight. For both sexes there was a systematic tendency for tall and heavy persons to underestimate their height and weight and conversely for small and thin persons to overestimate these measurements. The influence of controlled weight on self-reported height was statistically significant (women: P less than 0.05; men: P approximately 0.05,) and so was the influence of controlled height on self-reported weight (women: P less than 0.002; men: P less than 0.001). The reported heights and weights of women and men converged towards a pair of desired measures for each sex, ie the values of height and weight for which the reported and the controlled values were identical. For women the desired measures were 159.6 cm and 45.2 kg, for men 176.7 cm and 68.9 kg. The demonstrated differences between self-reported and controlled body measures may invalidate data and conclusions of insurance studies and population surveys based on self-reported measures.