Isolated haemosiderin contained iron and nitrogen in a weight ratio of 6.75, with phosphorus and no detectable haem. Considerably more iron was released from haemosiderin under acidic conditions than under neutral conditions in the presence of ascorbate, nitrilotriacetate or dithionite. Unlike the situation with ascorbate, chelators such as citrate, ADP or succinate induced the release of only some iron, with almost no pH-dependence. Dehydroascorbate (the oxidized form of ascorbate with no reducing capacity) behaved like citrate, ADP, succinate or desferal, rather than like ascorbate itself, in releasing iron. GSH had less effect on the release of iron than these chelators, but in the presence of a small amount of chelator the release of iron increased, especially under acidic conditions. Thus reduction, chelation and pH were all found to be important factors involved in the release of iron from haemosiderin. Investigation by e.p.r. of hydroxyl-radical production by the released iron showed high radical productivity at an acidic pH. However, at a physiological pH, almost no radical formation was detected, except in the presence of nitrilotriacetate. These findings suggested that, under physiological conditions, haemosiderin was not an effective iron donor and was almost not involved in radical production. Under acidic conditions, however, such as in inflammation, hypoxia and in a lysosomal milieu, it could possibly be an iron donor and is thought to be implicated in radical production and tissue damage in iron-overloaded conditions.