We aimed to investigate the risk factors of spontaneous osseous fusion (SOF) of the atlantoaxial joint after closed reduction under general anesthesia followed by halo fixation (remodeling therapy) for chronic atlantoaxial rotatory fixation, and to elucidate the recovery mechanism of the rotatory range of motion (ROM) after halo removal. Twelve patients who underwent remodeling therapy were retrospectively reviewed. Five patients with SOF were categorized as the fusion group and seven patients without SOF as the non-fusion group. Three dimensional CT was used to detect direct osseous contact (DOC) of facet joints before and during halo fixation, while dynamic CT at neutral and maximally rotated head positions was performed to measure rotatory ROM after halo removal. The duration from onset to initial visit was significantly longer (3.2 vs. 5.7 months, = 0.04), incidence of DOC during halo fixation was higher (0/7 [0%] vs. 4/5 [80%], = 0.004), and segmental rotatory ROM of Occiput/C1 (Oc/C1) at final follow-up was larger (9.8 vs. 20.1 degrees, = 0.003) in the fusion group. Long duration from the onset to the initial visit might induce irreversible damage to the articular surface of the affected facet, which was confirmed as DOC during halo fixation and resulted in SOF. Long duration from the onset to the initial visit and DOC during halo fixation could be used to suggest the risk for SOF. Nonetheless, rotatory ROM of Oc/C1 increased to compensate for SOF.