Approximately 0.9 percent of the white adult population of the United States and 1.1 percent of the adult population in Europe are affected by seropositive rheumatoid arthritis. As many as 10 percent of those patients may need an operation for atlantoaxial subluxation. Severe instability, especially when associated with vertical subluxation of the odontoid process, can result in progressive cervical myelopathy. Typically, occipitocervical fixation has been performed for these patients with use of autograft bone to achieve long-term stability through a solid fusion. Harvesting the bone graft increases the operative risk to the patient and may result in increased morbidity. In our experience, patients who have had no clear radiographic evidence of fusion following use of occipitocervical instrumentation seemed to have done as well as those who have had obvious fusion. One assumption is that the clinical improvement might be attributable simply to stabilization of the joint rather than to osseous fusion. A longitudinal study was performed on patients with rheumatoid arthritis who required an operation because of craniocervical or upper cervical instability. The results of clinical, radiographic, functional, and self-evaluations were studied to determine the efficacy of treatment and to compare the outcomes of bone-grafting with those of procedures done without bone-grafting in a group of 150 patients who underwent posterior occipitocervical stabilization with use of a contoured metal implant (a Ransford loop) that was affixed by sublaminar wires. Internal fixation was performed in 120 patients without bone-grafting and in thirty patients with use of autogenous bone-grafting. Preoperatively, 23 percent (thirty-five) of the 150 patients had mild neurological involvement (class II, according to the system of Ranawat et al.), 45 percent (sixty-eight) had objective findings of weakness and long-tract signs but were able to walk (class III-A), and 29 percent (forty-three) were quadriparetic and unable to walk (class III-B). The age of the patients at the time of the operation ranged from twelve to eighty-three years (mean, sixty-two years). There were significant improvements in postoperative Ranawat classes at all time-periods (range, p < 0.00005 to p = 0.0066) and in patient ratings of neck pain (range, p < 0.00005 to p = 0.0044) compared with preoperative scores. With the numbers available, there were no significant differences between the patients managed with a graft and those managed without grafting with respect to survival after the operation, Ranawat class, head or neck-pain rating, presence of subaxial abnormalities, radiographic craniovertebral motion, or vertical subluxation. Overall mortality at one month was 10 percent (fifteen of 150), although this value varied directly with the degree of preoperative disability. A second cervical spine operation was required in 11 percent (sixteen) of the 150 patients. While patients who have rheumatoid disease with anterior atlantoaxial subluxation should be treated with posterior atlantoaxial arthrodesis with use of bone-grafting and internal fixation, we believe that those who present with vertical instability and multi-level involvement can be treated with posterior occipitocervical stabilization with use of a contoured occipitocervical loop and sublaminar wire fixation without bone-grafting. Furthermore, we believe that the use of preoperative traction, bone cement, or a postoperative halo vest is unnecessary. Avoiding the harvesting of autogenous bone for grafting reduced the morbidity of this operation without compromising the outcome in these already sick patients.