Skeletal ages were assessed for 4,902 pairs of hand–wrist and knee radiographs of children aged 2–17 years. The FELS method was used to assess the hand–wrist, and the RWT method was used to asses the knee. These methods have the same conceptual and statistical basis. The mean absolute differences, within age- and sex-specific groups, ranged from 0.34 to 0.87 years. The SD of the differences ranged from 0.31 to 0.68 years and, like the means, tended to increase with age until about 8–11 years. The means and SD were generally larger for the boys than for the girls, except after 14 years. Within age- and sex-specific groups, the maximum absolute differences were from 1.45 to 2.99 years. These maximum differences changed irregularly with age but tended to increase until 11 years in the boys and 9 years in the girls. These large absolute differences between the skeletal ages of the hand–wrist and the knee cannot be explained completely by the effects of observer errors. The means of the relative (signed) differences between the skeletal ages of the hand–wrist and the knee were all close to zero, but there was a wide range within age- and sex-specific groups. These skeletal ages of the hand–wrist and the knee are not interchangeable. The 95th percentiles of the absolute differences show that, in at least 5% of children, the choice of the area for skeletal age assessment will markedly influence the evaluation of individuals. They also demonstrate that descriptions of populations based on distributions of skeletal ages from one part of the skeleton may be misleading.