Between 1982 and 1997, 403 consecutive patients (522 elbows) with rheumatoid arthritis underwent Souter-Strathclyde total elbow replacement. By the end of 2007, there had been 66 revisions for aseptic loosening in 60 patients. The mean time of follow-up was 10.6 years (0 to 25) The survival rates at five-, ten, 15 and 19 years were 96% (95%, confidence interval (CI) 95 to 98), 89% (95% CI 86 to 92), 83% (95% CI 78 to 87), and 77% (95% CI 69 to 85), respectively. The small and medium-sized short-stemmed primary humeral components had a 5.6-fold and 3.6-fold risk of revision for aseptic loosening respectively, compared to the medium-sized long-stemmed component. The small and medium-sized all-polyethylene ulnar components had respectively a 28.2-fold and 8.4-fold risk of revision for aseptic loosening, compared to the metal-backed ulnar components. The use of retentive ulnar components was not associated with an increased risk of aseptic loosening compared to non-retentive implants