Forty patients underwent 47 procedures for repair of a lumbar pseudarthrosis at the University of Minnesota, Twin Cities Scoliosis Center between 1973 and 1986. Forty-three procedures in 36 patients with a minimum follow-up of 2 years were reviewed. Symptoms at the time of surgery included back pain (100%), leg pain (61%), subjective neurologic symptoms (49%), and thigh pain (35%). The diagnosis was made most commonly on the basis of plain radiographs (70%). In 38 cases, posterolateral fusion was performed, using segmental sublaminar wiring in 37%, a variable screw plate in 29%>, Harrington compression rods in 16%, and no implant in 18%. Anterior interbody fusion was performed in six cases, including one combined fusion. Follow-up averaged 4.4 years. Solid fusion was achieved in 49% of cases. Eighty-six percent of patients continued to have low-back pain. Clinical outcomes were graded as excellent in 7%, good in 35%, fair in 9%, and failure in 49%. Significantly improved results were seen in patients in whom a solid fusion was obtained (p < 0.005), in patients who had undergone only one prior surgery on the lumbar spine (p < 0.01), and in patients in whom the initial fusion had been performed for spondylolisthesis rather than a primarily degenerative disorder (p < 0.025). A trend toward better results was seen in cases in which the patient did not complain of leg pain preoperatively, in which a Workers' Compensation claim was not involved, and in which a prior attempt at pseudarthrosis repair had not been performed. Repair of pseudarthrosis after lumbar spine fusion should be approached with caution because the results of repair are frequently disappointing. The outcome can be improved with careful patient selection and fusion technique.