One hundred radiographs of the left hand and wrist from 40 children with chronic renal insufficiency or end-stage renal disease were examined to determine which method of bone age estimation provided the most useful information in these children. The Tanner and Whitehouse method showed better repeatability than the Greulich and Pyle atlas or the Buckler handbook when a sample of the radiographs were assessed twice by the same observer. The Tanner and Whitehouse 20 (TW20) bone age showed less inter-observer bias than the radius, ulna and short bone age or the carpal bone age when three observers independently assessed the same sample of radiographs. TW20 was the most useful method of bone age assessment in this study of British children. An unexpected finding was that the carpal bones were significantly more retarded than the radius, ulna and short bones. Separate assessment of the carpal bone age may provide extra information of clinical relevance.