It is widely accepted that iron deposition in the iron storage protein ferritin in vitro involves Fe(II) oxidation, and that ferritin facilitates this oxidation at a ferroxidase site on the protein. However, these views have recently been questioned, with the protein ferroxidase activity instead being attributed to autoxidation from the buffer alone. Ligand exchange between another protein with ferroxidase activity and ferritin has been proposed as an alternative mechanism for iron incorporation into ferritin. In the present work, a pH stat apparatus is used to eliminate the influence of buffers on iron(II) oxidation. Here we show that the recent experiments questioning the ferroxidase activity of ferritin were flawed by inadequate pH control, that buffers actually retard rather than facilitate iron(II) oxidation, and that horse spleen ferritin has ferroxidase activity when measured under proper experimental conditions. Furthermore, high pH (7.0), a high Fe(II) concentration and the presence of Fe(III) all favour Fe(II) autoxidation in the presence or absence of ferritin.