Long term results of joint replacement are frequently expressed as survivorship; it therefore becomes increasingly important to have the ability to make valid comparisons between survivorship tables in order to determine accurately the statistical significance of published differences. We present a statistical method for direct comparison of survivorship tables. We have applied this technique to evaluate the survivorship of the unconstrained, uncemented Ring total knee replacement first implanted in 1975 and compare this to contemporaneous constrained prostheses. There is a significant improvement in survival, which suggests a beneficial long-term effect on the bone/implant interface of this changing design concept.