In a conservative treatment of lumbar disc herniation, authors customarily had included the flexion exercise untill 1991. Thereafter, the extension exercise started to be included for the selected patients and this study was designed to assess the clinical outcome of the extension exercise compared to the flexion exercise in the conservative treatments of lumbar disc herniations. 55 consecutive patients (31 males and 24 females having ages ranging from 19-68 years with a mean of 37.2) were included in this prospective study. Criteria for inclusion in this group were: 1. Contained herniations of a single lumbar disc, documented by CT or MRI; 2. no other concurrent spine pathology; 3. conservative treatments with an uniform program including the extension exercise; 4. follow-up for a minimum of one year. For comparison with this prospective group, another 62 consecutive patients (36 males and 26 females having ages ranging from 17-63 years with a mean of 35.7) were selected who were treated during 1991 with flexion exercise before this study was designed and who were matched with the designed criteria except for the direction of exercise. Apart from the therapeutic exercise, the conservative treatments also included medication, physiotheraphy, epidural injection, and back school in the both groups uniformly. The clinical outcome of the extension exercise group indicated that 28 (50.9%) patients excellent, 23 (41.8%) patients good, three (5.5%) patients fair, and one (1.8%) patient failed outcomes. In the flexion exercise group, there were 23 (37.1%) excellent, 27 (43.5%) good, seven (11.3%) fair, and five (8.1%) failed outcomes. From these results, it would seem to follow that the extension exercise group had superior clinical outcome compared to the flexion exercise, i.e. higher excellent and good outcomes (92.7% vs. 80.6%) and lower poor and failed outcomes (7.3% vs. 19.4%), respectively, Moreover, the excellent outcome in terms of full recovery without any pain and disability was more common in the extension exercise group (50.9% vs. 37.1%). A better clinical outcome was obtained in the extension exercise group of patients who were younger than 40 years and who had a history of three months or less compared with those who were older and had longer history of disease. The sizes of disc protrusion did not affect the clinical outcome. In conclusion, we would recommend that the extension exercise, instead of the flexion exercise, should be included in the conservative treatment of a contained herniation of lumbar disc for a better clinical outcome.