BACKGROUND Liver iron content (LIC) assessment by magnetic resonance imaging (MRI) is validated but not standardized. In a single center, we tried to assess the accuracy of a specific, simple MRI procedure adapted to high LIC from a well-established simple and routine procedure known to quantify LIC. METHODS In 27 cases of monthly transfused patients, we compared biochemical values of LIC assessed on liver biopsy specimens and results obtained by two signal intensity ratio of gradient echo imaging (R2*) MRI protocols. The first was Gandon's routine procedure previously validated in liver disease and the second, our own method, was an addition of a gradient echo sequence specifically adapted to high LIC encountered in hematology practice. RESULTS Twenty-seven liver biopsies were performed in 18 adult patients (myelodysplastic syndrome = 5, beta-thalassemia = 13). LIC by biopsy ranged from 1.4 to 54 mg/g liver dry weight (mg/g dw) (median 9.4 mg/g dw). Correlation between LIC by biopsy and by MRI with Gandon's procedure was good (R = 0.80) in patients with LIC falling within the range reported by Gandon. By contrast, a weak correlation was demonstrated (R = 0.52) in patients with high LIC (above 11.2 mg/g dw). With our sequences, the correlation was good both in the entire group of patients (R = 0.83) and in patients with LIC above 11.2 mg/g dw (R = 0.85). CONCLUSIONS Our results suggest that the addition of a specific shorter-gradient echo sequence to a very simple, fast technique produces an accurate estimation of LIC in post-transfusional iron overload.