The rapidity and duration of the response of non-transferrin-bound iron (NTBPI) to chelation therapy are largely unknown and have important implications for the design of optimal chelation regimens. Methodology was developed to measure simultaneously NTBPI, deferoxamine (DFO), and its major metabolite. NTBPI was present in all but 2 of 28 thalassaemia major (TM) patients who had received conventional subcutaneous DFO the previous night, suggesting a short duration of NTBPI clearance by DFO. The detailed kinetics of NTBPI were therefore studied in response to intravenous DFO at 50 mg/kg/27 h for 48 hours and compared in 17 regularly transfused TM and 8 untransfused thalassaemia intermedia (TI) patients to determine the influence of hypertransfusion and iron overload on NTBPI response. Before DFO infusion, NTBPI was present in all patients and was significantly higher in TI (4.52 +/- 0.53 mumol/L) than TM (2.92 +/- 0.03 mumol/L; P = .03). NTBPI values in TM correlated with transferrin saturation (r = .6, P = .03) but not with serum ferritin. Removal of NTBPI by intravenous DFO is in a biphasic manner. The initial rapid rate constant (alpha) was similar in TI (1.5 hour-1) and TM (1.6 hour-1), but the subsequent beta phase was slower (0.04 hour-1) in TI when compared with TM (0.4 hour-1, P = .002). Detectable NTBPI persisted during the beta phase, particularly in TI, despite an excess of plasma DFO also being present (steady state 8 mumol/L). On cessation of DFO infusion, NTBPI reappearance was rapid; the kinetics also being biphasic. The rapid initial rate constant (alpha = 2.5 hour-1) lasted less than 30 minutes and was approximately equal to the summation of the initial rate constant for removal of DFO (1.8 hour-1) and its major metabolite (0.6 hour-1). This was followed by a slower return to pretreatment levels, usually between 6 and 12 hours, which was faster in TI than in TM. This marked NTBPI lability supports the use of continuous rather than intermittent DFO in high risk patients.