To determine the efficacy of allograft bone in spinal surgery, a retrospective study was carried out on thirty-two patients who had had spinal fusion with banked bone from the femoral head between 1977 and 1983. The diagnoses were paralytic and neuromuscular scoliosis. At final follow-up (average, 3.7 years), the infection rate was 9.3 per cent. There were no pseudarthroses, and all patients had well marginated trabecular markings by eighteen months. The rates of complications and of fusion were comparable with or better than those obtained using autogenous bone graft.