OBJECTIVE We assessed the long-term (more than ten-year) outcomes of the Kudo type-5 elbow prosthesis in patients with rheumatoid arthritis (RA). METHODS We reviewed 41 elbows (Larsen Grade IV, n = 21; Grade V, n = 20) in 31 patients with RA who had undergone a Kudo type-5 total elbow arthroplasty (TEA) between 1994 and 2003, and had been followed up for more than ten years. The humeral component was cementless and the all-polyethylene ulnar component cemented in every patient. Clinical outcome was assessed using the Mayo elbow performance score. We calculated the revision rate and evaluated potential risk factors for revision. The duration of follow-up was a mean 141 months (120 to 203). RESULTS Aseptic loosening of the ulnar component occurred in 11 elbows. There was no radiolucency around any humeral component. There was one deep infection. The survival rate according to Kaplan-Meier survivorship analysis was 87.8% after five years and 70.7% after ten years. The range of extension/flexion was a mean -38° (-80° to 0°)/105° (30° to 150°) before surgery and -40° (-70° to -20°)/132° (100° to 150°) at the final follow-up, while the mean Mayo elbow performance score was 43 before surgery and 80 at final follow-up. Disease duration of RA up to the TEA of < 15 years and a pre-operative range of movement (ROM) of > 85° were significant risk factors for revision or aseptic loosening. CONCLUSIONS Although Kudo type-5 prostheses gave satisfactory results in the short-term, aseptic loosening increased after five years. In most cases, elbow function was maintained in the long-term without loosening of the implant. A short duration from the onset of RA to TEA and a large pre-operative ROM were significant risk factors for revision or aseptic loosening. Cite this article: Bone Joint J 2017;99-B:818-23.