DESIGN: A prospective, multicenter observational study. OBJECTIVES: 1) Identify correlates of the decision to perform arthrodesis in patients undergoing laminectomy for lumbar spinal stenosis. 2) Compare symptoms, walking capacity, and satisfaction 6 and 24 months after laminectomy alone and laminectomy with noninstrumented and with instrumented arthrodesis. BACKGROUND DATA: Few prospective studies have compared outcomes of laminectomy alone or laminectomy with noninstrumented or with instrumented arthrodesis in patients with degenerative lumbar spinal stenosis. There is uncertainty regarding the optimal use of arthrodesis and instrumentation. METHODS: Two hundred seventy--two patients undergoing--surgery for degenerative lumbar stenosis by eight surgeons at four centers were included in the study cohort. Of these, 37 had noninstrumented and 41 had instrumented arthrodesis. Logistic regression identified factors associated with arthrodesis. The principal outcomes-health status, walking capacity, back and leg pain, and satisfaction with surgery-were assessed 6 and 24 months postoperatively with univariate and multivariate techniques. Outcomes also were assessed in a restricted cohort of patients with at least 5 mm spondylolisthesis and/or 15 degrees scoliosis. Hospital costs were obtained from a computerized hospital cost accounting system. RESULTS: The major predictor of the decision to perform arthrodesis was the individual surgeon (P = 0.0001). Noninstrumented arthrodesis was associated with superior relief of low back pain at 6 months (P = 0.004) and 24 months (P = 0.01). This difference persisted in multivariate analyses, with borderline statistical significance. There were no significant differences in the other outcomes across treatment groups. Mean hospital costs of laminectomy alone and noninstrumented and instrumented arthrodesis were $12,615, $18,495, and $25,914, respectively (P = 0.0001). CONCLUSION: Findings were limited by the small number of participating surgeons, modest sample size that produced P values of borderline significance, and nonrandomized design. With these caveats in mind, the authors conclude: (1) The individual surgeon was a more important correlate of the decision to perform arthrodesis than clinical variables such as spondylolisthesis. (2) Noninstrumented arthrodesis resulted in superior relief of back pain after 6 and 24 months. (3) Instrumented arthrodesis was the most costly option. These results highlight the need for randomized controlled trials and cost effectiveness analyses of lumbar arthrodesis and instrumentation in patients with degenerative lumbar spinal stenosis.