The ability to quantitate transfusional iron overload is crucial for determining the need for and the efficacy of chelation therapy in patients with long-standing transfusion-dependent anaemias. We evaluated the usefulness of some indirect measures of iron overload in estimating the iron concentration in the liver--the most important iron storage organ--in 26 non-chelated adult non-thalassaemic patients. Liver iron concentration was determined non-invasively by magnetic resonance imaging (MRI). The standard error of the estimated liver iron concentration was 80 mumol Fe/g dried liver tissue when using the number of transfused blood units, and 93 mumol Fe/g when using a serum ferritin assay. Follow-up in 11 patients (12-48 months) revealed that serum ferritin is a poor measure of the liver iron concentration during iron chelation. However, this discrepancy was individually different and seemed to be dependent on the erythropoietic marrow activity. By monitoring the liver iron concentration by MRI, we compared the efficacy of chelation with desferrioxamine given either by subcutaneous continuous infusions or by bolus injections. Depletion of liver iron stores could be achieved efficiently by both regimens.