We assigned 67 patients with central lumbar stenosis alternately to either multiple laminotomy or total laminectomy. The protocol, however, allowed multiple laminotomy to be changed to total laminectomy if it was thought that the former procedure might not give adequate neural decompression. There were therefore three treatment groups: group I consisting of 26 patients submitted to multiple laminotomy; group II, 9 patients scheduled for laminotomy but submitted to laminectomy; and group III, 32 patients scheduled for, and submitted to, laminectomy. The mean follow-up was 3.7 years. Bilateral laminotomy at two or three levels required a longer mean operating time than total laminectomy at an equal number of levels. The mean blood loss at surgery and the clinical results did not differ in the three groups. The mean subjective improvement score for low back pain was higher in group I but there was also a higher incidence of neural complications in this group. No patient in group I had postoperative vertebral instability, whereas this occurred in three patients in groups II and III, who had lumbar scoliosis or degenerative spondylolisthesis preoperatively. Multiple laminotomy is recommended for all patients with developmental stenosis and for those with mild to moderate degenerative stenosis or degenerative spondylolisthesis. Total laminectomy is to be preferred for patients with severe degenerative stenosis or marked degenerative spondylolisthesis.