Bayley-Pinneau, Roche-Wainer-Thissen, and Tanner height predictions at various chronologic ages were compared with final adult height in 56 normal subjects and in 34 patients with abnormal growth pattern (11 with familial tall stature, 7 with idiopathic precicious puberty, 6 with Turner syndrome, and 10 with primordial small stature or Silver-Russell syndrome). The two recent methods (Roche-Wainer-Thissen and Tanner) gave very accurate results and were superior to the Bayley-Pinneau method in normal subjects and in patients with familial tall stature. However, they overestimated adult height grossly in precocious puberty and moderately in Turner syndrome and in primordial small stature. It is concluded that calculations based on coefficients and regression equations obtained from normal children (as in the Roche-Wainer-Thissen and Tanner methods) can only be used in normal children or in patients with normal growth potential under adequate treatment. Calculations based on percentages of adult height (as in the Bayley-Pinneau method) are preferable in conditions in which the growth potential in relation to bone maturation is inherently reduced and cannot be corrected by treatment.