The assessment of iron overload in transfusion dependent patients was studied by the new generation of 3 T imagers. From physical principles, a ratio of 2:1 could be expected between the transverse relaxation rates (R2*) at 3.0 and 1.5 Tesla. This was confirmed for liver and heart in former publications (1,2). We are investigating the suitability of a 3.0 T imager for iron measurements over the whole range of possible iron concentrations in the liver and other organs or glands. Here, we report our results on liver and spleen iron.Patients with transfusional siderosis (ß-thalassemia major n=8, age: 20-66 y) and one normal subject underwent a gradient recalled multi-echo (GRE) sequence (R2*). For assessment of transverse relaxation rates R2*, 2D single slice (n = 5-7, 8 mm data acquisition was performed at 1.5 T (Avanto®, Siemens AG, Erlangen, Germany) using a breath-hold GRE sequence (TR = 500 ms, 12 bipolar echoes with TE = 1.3-25.7 ms, Dt = 1.16 ms, flip angle = 20°, bandwidth 1955 Hz/pixel). At 3.0 T (Ingenia®, Philips AG, Eindhoven, Netherlands), a breath-hold 3D multi-slice (n = 20, 8 mm, oversampling 1.5) data acquisition was used (TR = 24.3 ms, TE = 1.2-23.05 ms, Dt = 20 · 1.15 ms, flip angle = 3°, bandwidth 1425 Hz/pixel). In patients with suspected severe liver iron burden (LIC > 2000 µg/gliver), an additional 3D sequence with shorter echo times was used (TE = 0.65-16.93 ms, Dt = 20 · 0.86 ms). In vivo liver iron concentration (LIC, dry-weight conversion factor = 6) was determined from hemosiderin/ferritin magnetic susceptibilities as noninvasively measured by SQUID biomagnetic liver susceptometry (BLS). R2* was determined from a mono-exponential fit to the echo-time dependent signal amplitudes (magnitude) averaged over a whole liver slice (large blood and bile vessels excluded) with constant signal level offset. In patients, liver iron ranged from LIC = 700 to 4200 µg/gliver. In the normal control, we found 176 µg/gliver (normal range: 100-400 µg/g). At 1.5 and 3.0 T, mean liver R2* rates averaged over a whole liver slice were 458 ± 307 s-1 and 719 ± 407 s-1, respectively. A non-linear relationship was found between R2*(3T) and R2*(1.5T) with R2*(3T) = (1514±237)·(1 – exp[-(1.7±0.5)·10-3 · R2*(1.5T)]) (r² = 0.95) due to the higher relaxivity at 3 Tesla, see Fig. 1. Below R2* < 1200 s-1 at 3 T equivalent to a LIC < 2400 µg/gliver (< 15 mg/gd.w.), the relationship could be linearized to R2*(3T) = (1.85±0.04)·R2*(1.5T) (r² = 0.99) with negligible intercept. These findings agree with similar findings from other reports (1,2). Spleen R2* was obtained between 30 and 1006 s-1 at 3.0 T with a linear regression coefficient of 2.17 related to 1.5 T (r² = 0.97). In order to overcome the saturation effect of the 3T relaxivity for severe liver iron overload one has to put constraints on the signal amplitude (spin density). This can be done by a shorter 1st echo time, an estimate of the signal amplitude from nearby surrogate tissue (subcutaneous thorax fat), or using the signal intensity ratio method (1). For liver iron measurements in severely overloaded patients with LIC > 2400 µg/gliver or > 15 mg/gdry weight, 1.5 Tesla imagers are better suited than 3.0 Tesla systems. For R2* rates below that level, 3.0 T systems with 3D data acquisition are more advantageous, especially for other organs and glands (pancreas, kidney, adrenals) due to a better image quality.