In order to develop new height prediction models for children with constitutionally tall stature, 55 such boys and 88 girls were recalled for measurement of adult final height (FH). Data on height (H), age (CA), and target height (TH) were collected from the hospital charts and radiographs of the left hand and wrist were retrieved and used for bone age (BA) determination [BA according to the methods of Greulich and Pyle (BAGP) and Tanner and Whitehouse (BARUS)]. Standard multiple regression techniques were used to develop prediction equations for FH. In addition to test the validity of the new equations, FH was measured in a second group of constitutionally tall children (n = 32) and compared with the predicted FH according to our models. In addition, a comparison was made with other prediction methods. Mean (SD) FH was 196.0 (4.9) cm in boys and 180.5 (3.8) cm in girls. The ultimate regression equation was FH (cm) = 216.07 + 0.75 x H + 0.25 x TH -11.09 BAGP + 0.74 x (CA x BAGP) for boys and FH = 161.42 + 0.73 x H + 0.15 x TH - 8.41 x CA -8.83 x BARUS -2.45 x M + 0.55 x (CA x BARUS) for girls. The models showed satisfying accuracy: the mean (SD) errors were -1.4(3.2) cm for boys and -0.5(3.1) cm for girls with corresponding mean (SD) absolute errors of 2.7 (2.2) cm and 2.0 (2.4) cm, respectively. Compared with the current prediction methods, the new models were quite promising. Their clinical validity has to be ascertained in larger groups of tall children.