Upper cervical cord area (UCCA) is strongly associated with physical disability in patients with multiple sclerosis, particularly in progressive stages of disease. Standard brain high-resolution 3D T1-weighted acquisitions that include the upper cervical cord can be used to provide estimates of UCCA. Depending on subject positioning in the scanner, gradient non-linearity can introduce up to 10% variability in UCCA measurements from volumetric brain MPRAGE scans. In planning a study, use of 3D correction methods provided by the scanner vendors seems optimal. Alternatively, to retrospectively correct data, we propose methods based on a phantom acquisition or normalization with vertebral body diameters.