Study Design. The cases of 47 patients with atlantoaxial dislocation exclusive of rheumatoid arthritis, cerebral palsy, and tumors as causative pathologies were reviewed after surgical treatment, which was performed between 1979 and 1993. Objectives. To investigate the surgical results of atlantoaxial dislocation itself without any systemic factors affected by rheumatoid arthritis, cerebral palsy, and tumors. Methods. Neck pain (or occipitalgia) and extent of myelopathy at follow-up evaluation were compared with that present before surgery. The results were classified into four groups: excellent (no pain or recovery rate in myelopathy of more than 50%), good (decreased pain or recovery rate of 25% to 50%), fair (no improvement of pain or recovery rate of zero to 25%), and poor (aggravation of pain or recovery rate less than zero). The average follow-up period was 4 years and 2 months. Results. Of the patients evaluated, 51% were assessed as excellent, 23% as good, 7% as fair, and 19% as poor. Pain relief was achieved in 95% of patients with non myelopathy. Extent of myelopathy, pathology of atlantoaxial dislocation (ligamentous or osseous instability), loss of reduction after surgery, and surgical procedures were recognized as the major factors affecting surgical results. Better surgical results were obtained in mild myelopathy cases (> 10 points in Japan Orthopaedic Association scoring), ligamentous instability, and cases without loss of reduction. The incidence of pseudarthrosis and loss of reduction was low in Brooks' method for atlantoaxial fusion and in Luque's segmental sublaminal wiring method for occipitocervical fusion. Conclusion. The best results occurred in patients with no myelopathy, and the worst results occurred in patients with severe myelopathy; therefore, surgery is best indicated for atlantoaxial dislocation with intractable pain or with mild myelopathy. To avoid pseudarthrosis and loss of reduction, a strong fixation method, such as Brooks' or Luque's segmental sublaminal wiring method, should be selected.