Between February 1985 and March 1990 44 patients with degenerative spondylolisthesis underwent primary surgery for spinal stenosis (in all cases the decompression preserved the facets bilaterally without discectomy) and were studied prospectively. Forty-three patients have been followed for > or = 2 years and are the basis of this study. There were three treatment groups: group I, no fusion (nine patients); group II, transverse process fusion with autogenous iliac bone graft without instrumentation (11 initial patients, with one lost to follow-up for a total of 10); and group III, transverse process fusion with autogenous iliac crest bone graft and instrumentation (24 patients, 18 with one-level pedicle fixation and six with two-level fixation). A higher proportion of group III subjects had a successful fusion compared with group II (p = 0.002). There was significantly more spondylolisthesis progression in groups I and II than in group III (p = 0.001). A higher proportion of "spondylolisthesis unchanged subjects" reported they were helped by the surgery than those whose spondylolisthesis progressed postoperatively (p < 0.01).