OBJECTIVES: To evaluate and compare nonoperative treatment methods for traumatic thoracic and lumbar compression fractures and burst fractures. DESIGN: Prospective randomized controlled trial with long-term follow-up. SETTING: Two general hospitals in the Netherlands. PATIENTS/PARTICIPANTS: Patients with a traumatic thoracic or lumbar spine fracture, without neurologic damage, with less than 50% loss of height of the anterior column and less than 30% reduction of the spinal canal were included. INTERVENTION: Patients in the compression group were randomized to physical therapy and postural instructions, a brace for 6 weeks, or a Plaster of Paris cast for 6 or 12 weeks. Patients in the burst group received a brace or a Plaster of Paris cast, both for 12 weeks. MAIN OUTCOME MEASUREMENTS: Follow-up examinations included radiographs, Visual Analogue Scores for toleration of treatment and persistent pain, and an Oswestry Disability Index at long-term follow-up. RESULTS: There were 133 patients: 108 in the compression group and 25 in the burst group. For compression fractures, physical therapy and brace were considered the most tolerable. Brace therapy scored significantly better on the Visual Analogue Scores for residual pain and on the Oswestry Disability Index. None of the treatments had any significant effect on the residual deformity measurements. For burst fractures, no significant differences were found. CONCLUSIONS: Brace treatment with supplementary physical therapy is the treatment of choice for patients with compression fractures of the thoracic and lumbar spine. Furthermore, more than 20% of all patients had moderate or severe back pain at long-term follow-up.