A retrospective review of 17 immature patients who underwent posterior spinal fusion of C1–C2 or C1–C3 was performed to establish the efficacy of the procedure, to determine the incidence of complications, and to identify those at high risk for complications. All had radiographic C1–C2 instability or dislocation. Etiologies included os odontoideum, fixed rotatory subluxation, atlantoaxial subluxation, type II dens fracture nonunion, and nonunion of a Hangman's fracture. Before surgery, 10 patients were neurologically intact, and seven had neurologic findings. Associated diagnosis included Down's syndrome in three and Ehler–Danhlos in one. Follow-up averaged 25 months. One neurologically intact patient had a C2 radiculopathy that resolved by 1 year. Three patients with Down's syndrome had complications: one with a preoperative Brown-Sequard syndrome had transient worsening in the immediate postoperative period, one with a preoperative myelopathy developed a late recurrence of a severe myelopathy that required odontectomy, and another sustained an intraoperative spinal cord contusion followed by postoperative quadriplegia and death due to respiratory failure. Of seven reporting neurologic symptoms before surgery, two had residual deficit in the late postoperative period. These two represented preventable technical errors. There were two nonunions, one of which required occiput to C2 fusion. In general, posterior spinal fusion of the upper cervical spine was found to be a reliable, safe, and predictable procedure, but extra caution should be employed when considering arthrodesis in patients with ongoing spinal cord compression, fixed dislocations, and inherited ligamentous laxity.