Quantitative susceptibility mapping (QSM) is a recent MRI technique that measures tissue magnetic susceptibility mostly influenced by iron, myelin and calcium content in the brain that may provide additional biomarkers [1-3]. QSM involves advanced numerical methods designed to solve the complex inverse problem from the measured magnetic field inhomogeneity (derived from phase) to the spatial distribution of susceptibility. The reconstruction algorithms now become sufficiently robust to be implemented in routine brain exams. We present here the installation of an online reconstruction solution and initial in vivo results at high 3 T and ultra-high 7 T clinical fields. Clinical brain QSM acquisition and automated processing at 3T and 7T for routine use MRI systems and sequences: Siemens Verio 3T and Magnetom 7T were used (VB17 revision) with 32 channel head coils. A 3D multi-echo gradient echo sequence was applied. At 3T, the R2* mapping protocol designed for multicenter studies [4] was used (1 mm resolution, Tacq=10min); at 7T, isotropic 600 µm resolution were targeted (Tacq=12.5min). A modified reconstruction software (ICE functor) was used to generate proper combined phase images for QSM in which the first-echo individual coil images are used as a reference to solve the known issue in coil combination [5]. QSM reconstruction: The amplitude and phase DICOM images were sent to a separate DICOM server based on dicom toolkit (dcmtk open source software). Additional scripts were automatically triggered on reception to start the QSM post-processing. QSM reconstruction involved a field inhomogeneity calculation and unwrapping step from the multiple echoes, followed by a brain extraction (BET) step, an estimation of the internal field and the final MEDI reconstruction [2]. The BET algorithm was modified at 7T for an enhanced brain coverage. Images were then sent automatically back to the MRI when completed. Methods Conclusion