To assess the reproducibility of a qualitative rating scale of cerebral atrophy on MRI in an aged population of demented and non-demented individuals, 4 raters independently judged cerebral atrophy (CA) in 13 regions on a 0-3 scale on 75 MRI scans, on two occasions. The level of agreement was expressed by kappa statistics as well as by analysis of variance for interexaminer reproducibility studies. The mean CA scores ranged from 13.3 to 21.2, indicating moderately high prevalence of cerebral atrophy in this sample. Complete agreement among observers was reached in 49.7% of the ratings during the first assessment and in 44.1% in the second assessment. The inter-observer agreement was poor in both sessions (mean overall kappas 0.34 and 0.24). The intra-observer agreement was moderate to good for all raters (mean overall kappa: 0.58). The average standard deviation of the differences between CA scores of the 4 raters was 5.12 in the first and 5.13 in the second session. The average standard deviation for the differences within raters was 2.4. We conclude that the assessment of cerebral atrophy using this rating scale on MRI in an aged population has a poor reproducibility among raters. However, it provides regional atrophy measurements and is quite reliable when performed by a single rater.