The evidence for the diagnostic accuracy of the four main imaging modalities used in low back pain (plain radiographs, CT, MR imaging, radionuclide bone scans) is variable in quality and limits the ultimate conclusions regarding the effectiveness and cost-effectiveness of diagnostic strategies. In addition, the frequent finding of abnormalities in normal adults limits the specificity of all of these tests. Nevertheless, MR imaging is likely in most cases to offer the greatest sensitivity and specificity for systemic diseases, and its performance is superior to that of radiographs and comparable with CT and radionuclide bone scans for most conditions causing neurologic compromise.