1. Urinary iron excretion after desferrioxamine has been examined in nine patinets with different iron-loading anaemias. Particular attention has been paid to individual variation in response and the kinetics of iron removal in order to determine the most efficient and convenient method of administration. 2. Twelve-hour subcutaneous infusions of desferrioxamine were comparable with intravenous infusions and gave a mean value of 62% more iron excretion than similar intramuscular bolus doses (range 20--125%). 3. Increasing doses as 12 h subcutaneous infusions produced a linear increase in iron excretion, which was followed by a tendency to reach a plateau. Iron excretion varied greatly between patients, was not related solely to age or estimated iron load, and in most cases was increased by ascorbic acid saturation. 4. Maximum iron-excretion rates were achieved after 3--6 and then maintained throughout an infusion. With bolus injections excretion rates declined rapidly after the first 6 h, during which approximately 60% of the total iron excretion occurred. 5. The dose and method of administration should be 'tailor-made' for each patient. Overnight 12 h subcutaneous infusions can be both as effective as similar doses given over 24 h and a practical way of achieving substantial negative iron balance. 6. Since children receiving regular blood transfusions for congenital anaemias such as thalassaemia usually die at the end of the second decade, this approach to iron chelation offers the possibility of alleviating what have hitherto been fatal-iron loading states.