In todays Radiotherapy Treatment Planning, RTP, it is common to use Computed Tomography, CT, together with Magnetic Resonance Imaging, MRI, where CT provides electron density information and a geometrical reference, and where MRI provides superior soft tissue contrast. To sim- plify the workflow and improve treatment accuracy, research groups have demonstrated how to exclude CT and use a MRI-only approach. In this thesis, a method for spatial distortion analysis, ultimately enabling quality assurance, QA, of the spatial accuracy of MRI, was defined, tested and evaluated. A phantom was built to cover the entire clinical Field Of View, FOV, and 6mm-diameter fluid filled paintball markers were placed in a well-defined geometrical pattern within the phantom, and used as positive contrast. The phantom was imaged with a 3D Fast Gradient Echo sequence and a 3D Fast Spin Echo sequence. The markers were identified in the image data by a MATLAB-algorithm, and the location of the center of mass was calculated for each marker and compared to a theoretical reference. The location error was defined as the spatial distortion - a measurement of the spatial accuracy. Imaging parameters were altered and the effect on the spatial accuracy was analyzed. The spatial distortions were successfully measured within the entire (maximal) clinical FOV. It was shown that high readout bandwidth reduced distortions in the frequency encoding direction. These distortions could thus be attributed to B0-inhomogeneities. It was also determined that increasing the readout bandwidth to the maximum value reduced the maximum distortions in the frequency encoding direction to the same level as the maximum distortions in the other two phase-encoding directions of the 3D acquisitions. The voxel size had a very small effect on the spatial accuracy, enabling large voxelsize to be used when imaging the phantom, to decrease the scan time. The method was deemed capable of serving as a basis for QA of the spatial accuracy of large FOV MRI, which is needed in future MRI-only RTP approaches.