Object: aim of this study was to analyse the observer variability in the diagnosis and definition of disc pathologies with low and high-field strength MR scanners. Material and methods: 95 patients with low back pain or radicular pain who were referred from two different centers were included in the study. Fifty-seven patients were scanned with 0.3 T MR (group 1) and 38 patients with 1.5 T (group 2). The intraobserver and interobserver reliability were assessed with the cappa coefficient which was characterised as follows: values less than 0.0 = 'poor' agreement, values 0.01-0.2 = 'slight' agreement beyond chance, 0.21-0.4 = 'fair' agreement, 0.41-0.60 = 'moderate' agreement, 0.61-0.80 = 'substantial' agreement and 0.81-1.00 = 'almost perfect' agreement. Results: intraobserver agreement in group 1 and group 2 for both readers was 'almost perfect' in differentiating normal and pathological discs; 'substantial-almost perfect' in defining the disc pathologies, 'moderate-substantial' in root compression, and 'moderate-substantial' in spinal stenosis. Interobserver agreement was 'almost perfect' in differentiating normal and pathological discs, 'substantial' in defining disc pathologies 'moderate' in root compression and, 'moderate' in spinal stenosis in the group 1, whereas in group 2, it was 'almost perfect' in differentiating normal and pathological discs, 'almost perfect' in defining disc pathologies, 'slight-substantial' in root compression and 'moderate' in spinal stenosis. Conclusion: in the diagnosis of root compression and spinal stenosis, the intra and interobserver agreements were relatively poor with both high and low-strength field MRIs, indicating a need for more objective criteria. In differentiating normal and pathologic appearance of disc, the interobserver agreement was considerably better with high-field compared to low-field strength MRI. In cases where this definition is important, high-field strength scanners should be preferred. (C) 2003 Elsevier Ireland Ltd. All rights reserved.