In the retrospective study reported here, we compared the longitudinal growth in three groups of children with thalassemia major who received a similar transfusion program but different schedules of chelation treatment. In those patients who initiated deferoxamine (DF) administration by daily subcutaneous infusion (50 to 80 mg/kg/day) simultaneously with the beginning of transfusion (at 8 +/- 6 months), mean height at 2 to 6 years of age was significantly reduced in comparison (1) with those patients who initiated DF subcutaneous treatment after 3 years at similar doses and (2) with those who were treated intramuscularly with small doses. In the patients treated at an early stage, those with more marked stunted growth had a clinical and radiologic ricketslike syndrome associated with joint stiffness. Mineral metabolism studies in these patients showed a reduction of hair and leukocyte zinc levels and leukocyte alkaline phosphatase activity. Our findings indicate that DF administration at high doses by continuous infusion before iron overload has been established adversely affects longitudinal growth. By contrast, after 3 years of age, even large doses (in the order of 100/mg/kg/day) did not result in growth retardation. The growth retardation observed may be related to chelation of other trace elements, including zinc, in the presence of low iron burden, to the direct toxic effect of unchelated DF by interference with critical iron-dependent enzymes, or both. These results indicate that in patients with thalassemia major, DF administration should be initiated only after iron accumulation is established, namely, around 3 years of age, after 20 to 30 transfusions, which are usually associated with ferritin levels in the range of 800 to 1000 ng/ml. At this age, deferoxamine doses should be established on the basis of iron balance studies and dose response curves. Doses higher than 50 to 60 mg/kg do not adversely affect growth but produce toxic side effects on acoustic and visual pathways and therefore should not be used. Longitudinal growth monitoring of DF-treated patients is warranted.