All patients referred for orbital imaging to the neuroradiology department of the Institute of Neurological Sciences in Glasgow over a three year period were enrolled in the study and were scheduled to undergo both magnetic resonance imaging and computed tomography. A total of 101 of the 110 referred patients were deemed suitable for analysis. Details of key presenting symptoms, signs, and a pre-imaging diagnosis were recorded prospectively. A final diagnosis was obtained by histology in 65% of cases with an orbital abnormality, by a minimum of one year of clinical review in 19.5%, by response to antibiotic or steroid therapy in 8.5%, or by conclusive investigations such as carotid angiography in in 7% of patients, 29% of the patients had no detectable orbital disease despite a minimum one years' follow-up, and so were regarded as a "normal" group. The images were interpreted prospectively by separate masked observers. The diagnostic accuracies of the two techniques were compared to the final diagnosis. The two imaging methods were shown to be comparable in overall diagnostic accuracy, with a small and statistically non-significant advantage held by magnetic resonance imaging. Interpretation of the two investigations gave more accurate information in different types of disease.