The results of lumbar myelography and computed tomography (CT) were compared in 51 consecutive patients with clinically suggested lumbar disc herniation. A total of 100 intervertebral levels were examined. At 62 levels, either L4/L5 or L5/S1, myelography was normal. CT showed no pathologic changes at 55 levels. The results concurred between myelography and CT in 89 per cent of the patients with normal findings. Four cases of disc herniation and one bulging disc, which had been missed at myelography because of a large epidural space at L5/S1, were picked up by CT. Two of these were verified by surgery and two were treated conservatively. There was one possible false negative case with CT as well. Abnormalities were shown at 38 intervertebral levels, 22 in the bulging and 16 in the herniated disc group. The pathologic changes concurred in 84 per cent between the two investigations. For intervertebral disc herniation the true negative rate was, for myelography 88 per cent and for CT 97 per cent. The treatment strategy could have been based on CT alone at 37 out of 38 levels (97%), and on myelography alone at 34 out of 38 levels (89%). Furthermore, at CT the imaging of soft tissues and intervertebral joints was superior to that at myelography. It is concluded that CT should be the primary examination method of imaging for lumbar disc herniation. Myelography is, however, to be preferred where the level of the lesion is clinically unclear or when the entire lumbar region and thoraco-lumbar junction are to be examined.