Forty-two consecutive patients with suspected lumbar disc protrusions were studied prospectively to compare the diagnostic accuracy of low-field strength MRI with that of radiculography. Thirty patients subsequently underwent surgical exploration at 33 levels and the operative findings were compared to both methods of diagnostic imaging. All patients had MRI studies, whilst 29 patients also had radiculography. Both studies were evaluated without prior knowledge of the clinical signs, operative findings or the results of other imaging techniques. MRI predicted the correct diagnosis in 29 of the 33 levels explored, an accuracy of 88%. All discs proven to be abnormal demonstrated a reduced signal from the nucleus pulposus. There were two false positive results and two doubtful cases but no false negatives. Radiculography predicted the correct diagnosis in 24 of 32 levels explored, an accuracy of 75%. There were two false positive results, five doubtfuls and one false negative. The overall accuracy when both tests were considered rose to 94%. Of the remaining 12 patients all except one showed good correlation between the MRI and radiculographic findings. These results indicate that low field strength MRI is slightly better than radiculography in diagnosing lumbar disc protrusions.