The accuracy of computed tomography (CT), myelography, CT-myelography (myelo-CT) and magnetic resonance imaging (MRI) for the diagnosis of lumbar herniated nucleus pulposus (HNP) is compared prospectively in 59 patients, all of whom underwent surgical exploration. All tests were read independently of each other and the level of confidence in each diagnosis was recorded. The results are based on the negative (61) as well as positive (59) findings at the 120 disc sites (level and side) explored. Magnetic resonance imaging was the most accurate test (76.5%) compared with myelo-CT (76.0%), CT (73.6%), and myelography (71.4%). The false positive rate was lowest for MRI (13.5%) followed by myelography (13.7%), CT (13.8%), and myelo-CT (21.1%). The false negative rate was lowest for myelo-CT (27.2%) followed by MRI (35.7%), CT (40.2%), and myelography (44.1%). In that subset of 19 patients who had prior surgery, myelography was the most accurate means of diagnosing lumbar HNP (88.8%), followed by MRI (83.3%), myelo-CT (78.4%), and CT (72.6%). The false positive rates in these patients were 11.6% for myelography, 13.2% for MRI, 14.5% for CT, and 16.4% for myelo-CT; the false negative rates were 22.7% for MRI, 24.4% for myelography, 29.5% for myelo-CT, and 47.7% for CT. Magnetic resonance imaging compares very favorably with other currently available imaging modalities for diagnosing lumbar HNP. Magnetic resonance imaging is painless, has no known side effects or morbidity, no radiation exposure, and is noninvasive. The authors recommend it as the procedure of choice for the diagnosis of most lumbar disc herniations.