Longitudinal monitoring of young sportspersons (50 males and 35 females) was used to verify the accuracy of two methods (Bayley, 1959; Prokopec, 1979) of adult body height prediction based on a knowledge of the skeletal age of the subjects. The methods are also compared with prediction based on calendar age. The mean difference in prediction was 2% of adult body height using the Bayley (1959) method and 1.5% using the method of Prokopecet al.(1979), similarly for males and females. The categories of deviations 0‐2 and 2‐5 cm included 96% of males and 100% of females according to the Prokopec method, whereas in the case of Bayley's the figures were 80% and 100% respectively. Thus the percentile growth graph of Prokopecet al.seems more suitable, particularly in the case of above‐average height.