In addition to its growth promoting effect, GH has profound metabolic effects that have not always been evaluated in longitudinal studies. We have recently shown that the effect of GH on body composition can be evaluated by magnetic resonance imaging measurement of adipose and muscle tissue cross-sectional (cs) areas in the thigh. The aim of this study was to evaluate the long-term effects of human GH (hGH) (0.2 IU/kg day) on muscle and adipose tissue mass during a 3-yr treatment period and after 1 year's withdrawal in short SGA (small for gestational age) children. Measurement of muscle and fat tissue mass by magnetic resonance imaging of the thighs was used to study the metabolic effect of hGH in 14 prepubertal short children born SGA. Results were compared with those of a control group of 7 normal children followed longitudinally. An increase of muscle tissue cs area was observed during the 3 yr of hGH treatment, an increase which was significantly different during the first 2 yr of treatment from that seen in controls (+31.2+/-2.6% and +18.1+/-1.8% during the 1st and 2nd year, respectively, vs. +9.1+/-2.6% change during 1 yr in controls). After a significant decrease in adipose tissue cs area during the first year of therapy (-16.4+/-3.4% vs. baseline values), an increase in adipose tissue cs area occurred during the second and third years. At the end of the third year, the muscle tissue cs area change was significantly greater in SGA-treated children, as compared with controls (+71.6+/-4.6% vs. 22.1+/-4.6%; P < 0.001), whereas the adipose tissue cs area change was similar in the two groups (+12.6+/-9.5% vs. +19.9+/-4.2%). After hGH withdrawal, the effects were opposite after 3 months, as compared with those observed after the first 3 months of hGH administration, whereas no additional significant change was seen after 1 yr off treatment, indicating the maintenance of muscle and adipose tissue mass. In conclusion, hGH administered to SGA children is effective in improving growth velocity and has long-term effects on muscle and adipose tissue mass. These effects may lead to speculation about the sensitivity of these tissues to GH. The physiological consequences of such effects must be evaluated.