The clinical utility of dual sequence (T1- and T2-weighted) magnetic resonance (MR) imaging in estimating liver iron concentration (LIC) in 32 transfusion-dependent beta-thalassaemia major (24 females; age 18.5+/-5.9 years) patients on desferrioxamine was evaluated. Signal intensity ratios (SIR) between liver, spleen and pancreas to psoas muscle were determined on both sequences. Relationships between clinical and MR parameters, and accuracy of SIR thresholds in determining adequacy of chelation from LIC were analysed. Liver T1- and T2-SIR were related to LIC (P < 0.001). T1-SIR < 0.60 predicted severe iron overload (LIC > 15 mg/g) with sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of 100%, 87%, 33% and 100% respectively. T2-SIR < 0.1 yielded 100% sensitivity, 93% specificity, 50% PPV and 100% NPV. T1-SIR > or = 1.1 predicted LIC < 7 mg/g with 69% sensitivity, 88% specificity, 85% PPV and 74% NPV. T2-SIR > or = 0.20 yielded 56.5% sensitivity, 94% specificity, 90% PPV and 71% NPV. LIC correlated with liver T1-SIR, liver T2-SIR and serum ferritin (r = -0.76, -0.65, 0.47, respectively; P < 0.01). Serum ferritin was inversely related to liver T1-SIR, liver T2-SIR and spleen T2-SIR (r = -0.35, -0.43, -0.40, respectively; P < 0.05). Mean total transfusion burden was not related to any MR parameter. Although neither MR sequence was a highly accurate predictor of LIC, SIR thresholds are useful to determine presence of iron overload and adequacy of chelation treatment.