Radiculography is used a great deal in our region and is often the only method available for examination of the lumbar spine. However, there has been little evaluation of the effectiveness of radiculography for lumbar diseases. To determine its value and indications in lumbar diseases we studied 322 patients who underwent radiculography. We studied 223 patients retrospectively and 99 patients prospectively. Radiculography had a sensitivity of 100%, a specificity of 86%, a positive predictive value of 92.7% and a negative predictive value of 100%. We found that the diagnostic accuracy of radiculography was higher if the examination was requested by a neuropathologist, if the patient had been ill for more than 2 months or had cauda equina syndrome, acute root pain, paralysis or debilitation. Such precise indications make it possible to avoid excessive examination and to make the best use of radiculography, particularly in areas without CT and MRI facilities. However, the examination should always be performed by a neuroradiologist or a physician with several years' radiculography experience.