Nineteen children and adolescents receiving repeated transfusions and subcutaneous desferrioxamine treatment were investigated in an attempt to quantitate iron overload non-invasively. Before patients were started on desferrioxamine individual relationships were correlated for 12 to 36 months between transfused iron, absorbed iron estimated gastrointestinally, and increasing serum ferritin concentrations. Patients with inflammation, increased liver enzymes, or haemolysis were excluded from analysis. The relationship between the variables could be described by a logarithmic regression curve (y = transfused iron [plus eventually gastrointestinally absorbed iron] = iron overload = a+b log [x = serum ferritin]) for each individual patient. All patients showed close correlation (R2) between x and y (median R2 of 0.909, 0.98, and 0.92 in thalassaemia, aplastic anaemia, and sickle cell anaemia patients, respectively). When started on desferrioxamine, current serum ferritin concentrations were used to derive the iron overload from each individual regression curve. The derived estimated iron overload ranged from 0.6 g to 31 g. Left ventricular dilatation was observed in three patients with beta thalassaemia and in one patient with aplastic anaemia with median iron overload of 20.7 (14.1-31.3) g and 24.0 g respectively. Hypothyroidism was found in four patients with beta thalassaemia and one patient with aplastic anaemia with iron overload between 14.7 (6.8 and 26.1) g and 15.1 g respectively. Human growth hormone deficiency was detected in three patients with beta thalassaemia with an iron overload of 4.2 (3.5-6.8) g; all three patients had excellent desferrioxamine compliance.