The bioavailability of ferritin iron was evaluated in human subjects using radiolabelled [55Fe]ferritin isolated from bovine spleen and liver. Preliminary studies with bovine spleen ferritin labelled in vitro demonstrated an inappropriately high absorption compared with ferritin labelled in vivo, and the latter was therefore used in all subsequent absorption studies. In 10 subjects, geometric mean absorption from 5 mg of ferritin iron was 3.8% when taken without and 3.2% when taken with food (P > 0.05). These values were significantly lower than absorption from the same dose of iron given as ferrous sulphate, which averaged 24.1% without and 8.2% with food. When the iron dose was increased 10-fold, absorption of ferritin iron averaged only 0.6-0.7% with or without food as compared with 7.9% without and 2.6% with food when the iron was given as ferrous sulphate. In a further study, mean absorption from bovine spleen ferritin of 4.0% did not differ significantly from the mean of 2.7% observed with bovine liver ferritin. These findings confirm previous studies indicating that ferritin iron is poorly absorbed. Furthermore, its use as a pharmaceutical iron preparation cannot be advocated.