The aim of this study is to assess the accuracy of MRI alone in the differentiation of soft cervical disc protrusion from osteophytic compression in cervical disc disease. In a retrospective study, the MRI scans of 41 patients with cervical disc disease, who had previously undergone surgery, were presented to three independent observers, randomly on two different occasions, to identify the accuracy of the diagnosis of the presence of hard or soft disc or both as a cause of compression. The observers (two neurosurgeons and one neuroradiologist) were not involved with the treatment of the cases at any stage and were unaware of the surgical findings. Their observations were compared with those of the surgeon recorded at operation. The intra-observer agreement was poor for diagnosis into three categories as hard or soft disc or both. In distinguishing between the presence or absence of hard disc, there was moderate to good (Kappa = 0.6) intra observer and fair to moderate (Kappa = 0.4) interobserver agreement. The sensitivity of diagnosis of a hard disc was high (87%) but specificity was low (44%), due to the overestimation of the presence of hard disc. There was a significantly higher incidence of hard disc in the elderly age group (76% over the fifth decade, P = 0.0073). It is concluded that MRI alone is not a very efficient diagnostic tool in distinguishing between hard and soft disc in the cervical disc disease.