Background: In multiple sclerosis (MS), recent work suggests that cervical cord atrophy is more consistently correlated with physical disability than brain white matter lesion load and atrophy. Although spinal cord imaging has not been routinely obtained in many clinical trial and research studies, brain volumetric imaging usually has and includes the upper cervical cord. Objectives: Using volumetric T1-weighted brain images, we investigated cross-sectional area measures in the uppermost cervical cord and compared them with areas at the standard C2/3 level. Methods: Using T1-weighted brain scans from 13 controls and 37 people with MS, and an active surface technique, cross-sectional area was measured over 5 mm and 1 mm cord segments at C2/3, below the level of odontoid peg, and 2 cm and 2.5 cm below the pons. Brain volume was also measured. Results: Cord area measurements were most reliable in a 5 mm segment 2.5 cm below the pons (inter-rater coefficient of variation 1.5%, intraclass correlation coefficient 0.99). Cord area at this level correlated more with that at C2/3 area than with brain volume (r=0.811 with C2/3, r=0.502 with brain volume). Conclusion: Whereas the standard C2/3 level is often not within the field of view on brain images, the level 2.5 cm below the pons usually is, and measurement at this level may be a good way to investigate upper cervical cord atrophy when only brain images are available. (C) 2014 Elsevier B.V. All rights reserved.