The goal of this prospective study was to assess whether intensive physical training during puberty could alter the growth potential of adolescent female athletes. Height, sitting height, leg length, weight, body fat, and pubertal stage of 22 gymnasts aged 12.3 +/- 0.2 years (mean +/- SEM), with an average training period of 22 hr/wk, and of 21 swimmers aged 12.3 +/- 0.3 years (average training period 8 hr/wk) were recorded half-yearly for a mean period of 2.35 years (range 2.0 to 3.7 years). Adult height predictions were performed with the methods of Bayley and Pinneau; Roche, Wainer, and Thissen, and Tanner et al. Growth velocity of gymnasts was significantly lower than that of swimmers from 11 to 13 years of bone age (p < 0.05), with a mean peak height velocity of 5.48 +/- 0.32 cm/yr versus 8.0 +/- 0.50 cm/yr for swimmers. Height standard deviation score decreased significantly in gymnasts with time (r = -0.747; p < 0.001). This observation was not associated with a significant alteration of chronologic age/bone age ratio. By contrast, height standard deviation score remained unchanged in swimmers (r = -0.165; p = 0.1). A marked stunting of leg-length growth was observed in gymnasts from 12 years of bone age, resulting in a marked difference in overall sitting-height/leg-length ratio (gymnasts 1.054 +/- 0.005 vs swimmers 1.100 +/- 0.005; p < 0.001). Concomitantly, predicted height of gymnasts decreased significantly with time (Tanner et al.: r = 0.63, p < 0.001; Bayley-Pinneau: r = 0.44, p < 0.001), whereas those of swimmers did not change. We conclude that heavy training in gymnastics (> 18 hr/wk), starting before puberty and maintained throughout puberty, can alter growth rate to such an extent that full adult height will not be reached. The mechanisms underlying these observations are not settled; we suggest that prolonged inhibition of the hypothalamic-pituitary-gonadal axis by exercise, together with or because of the metabolic effects of dieting, is responsible for them.