Several methods for adult height prediction are currently in use. All are subject to a wide range of error which is thought to result, at least in part, from the use of bone age estimation. Following the suggestion made by Karlberg to predict adult height of pubertal children by the use of the 'Infancy-Childhood-Puberty model' (ICP), growth data of 39 normal boys who were followed from infancy until adult height was attained were reviewed. Use of the ICP model alone and without bone age resulted in more accurate predictions of adult height than those made by the methods which required bone age determination, the Bayley-Pinneu (BP) and the Tanner-Whitehouse methods (TW). The absolute error of prediction was 3.4 cm as compared to 5.3 (BP) and 4.9 cm (TW) (p < 0.05) and maximal range of error was 10 cm as compared to 22 (BP) and 21 cm (TW) (p < 0.05). Finally, based on the ICP model, a new equation which incorporates paternal height was been derived. This equation, termed ICP-New (ICPN), resulted in even better accuracy: absolute error of prediction was 2.3 cm as compared to 3.4 cm (ICP) and maximal range of error was 6 cm as compared to 10 cm (ICP) (p < 0.05). It is concluded that the ICP model and the ICPN equation may predict the adult height of pubertal boys more accurately than the methods which use bone age.