Atlantoaxial joint is a possible affected site during rheumatoid arthritis (RA) and, in this work, we evaluated its occurrence and associated characteristics in a “real-life” cohort. By a medical records review study of RA patients longitudinally followed-up, the occurrence of severe atlantoaxial joint involvement was estimated (incidence proportion and incidence rate per 1000 person-years at risk). Regression analyses were also exploited to evaluate possible associated factors. Based on these findings, a prospective recruitment was performed to build a descriptive cross-sectional study in evaluating a subclinical atlantoaxial joint involvement in patients with the same clinical characteristics. Retrospectively, 717 patients (female 56.6%, age 64.7 ± 12.3 years) were studied. The incidence proportion of severe atlantoaxial joint involvement was 2.1% [1.5–2.5], occurring in 15 out of 717 patients, and identified by both MRI and CT scan. Considering over 3091 person-years, an incidence rate of 5.2 × 1000 [2.9–8.3] person-years was estimated. Regression analyses suggested that male gender, a longer disease duration, ACPA positivity and extra-articular manifestations resulted to be significantly associated with a severe atlantoaxial joint involvement. Given these findings, 30 asymptomatic patients were selected according to these clinical characteristics and underwent MRI of cervical spine. To date, almost 50% of these asymptomatic patients showed a subclinical atlantoaxial joint involvement. The occurrence of the severe atlantoaxial joint involvement in RA patients was estimated in a “real-life” setting. Male gender, ACPA positivity, long disease duration, and extra-articular manifestations could be associated with the severe atlantoaxial joint involvement in RA. MRI could provide a useful clinical tool to early evaluate the atlantoaxial joint involvement in RA, also in asymptomatic patients.