The goals of this study were to test the hypothesis that girls with Turner syndrome (TS) experience growth failure early in life and to establish model-based normative growth charts for 0- to 8-year-old American girls with TS. Full-term girls with TS who had 5 or more measurements of height obtained during their first 10 years of life prior to initiation of growth hormone, estrogen and/or androgen therapy were eligible for this study. A nonlinear mixed-effects model comprising the first two components of the infancy-childhood-puberty (ICP) model of growth was fitted to the longitudinal height measurements and compared with those of healthy American girls. Height measurements (n = 1,146) from 112 girls with TS (45,X: 57.1%; 45,X/46,XX: 12.5%; 46,X, iso(X): 4.5%, and other: 25.9%) were analyzed. Mean height SDS fell from –0.68 at birth to –1.60 at 1 year, –1.80 at 2 years and –1.95 at 3 years. When compared to controls (676 girls, 4,537 measurements), girls with TS grew more slowly due to three principal factors: a slow growth rate of the infancy component, a slow growth rate at the onset of the childhood component, and delayed onset of the childhood component. Traditional concepts of growth failure in TS should be revised. Physicians should consider the diagnosis of TS in any girl with unexplained failure to thrive or short stature, even in the first 3 years of life.