The efficacy of 'limited posterior surgery' for metastases in the thoracic and lumbar spine was studied prospectively in 51 patients (32 men and 19 women, mean age 64 years). The most common primary tumors were prostate, breast, and renal carcinoma, 37 patients had metastases in the thoracic spine and 14 in the lumbar spine. Indications for surgery were severe pain or neurologic deficit. Of the 46 patients with neurologic symptoms, 25 were unable to walk. Surgery was confined to direct or indirect decompression and stabilization with a pedicle screw fixator over few segments as possible. Pain, as well as a variety of functional performance parameters and residential status were registered preoperatively and after surgery at 3, 6, 9, and 12 months, and at 6-monthly intervals thereafter. Pain was rated by the patient on a Visual Analog Scale, and functional performance was assessed with the Eastern Co-operative Oncology Group (ECOG) Performance Status Scale. We had no perioperative neurologic deterioration or death. Nineteen of the 25 nonambulatory patients regained their walking ability. Postoperative pain relief was significant and lasting over time. Nearly half of the patients attained improvement in functional performance. The median survival was 8 months. Older age and intact postoperative walking ability were positive factors for survival.