With rare exception, the bulk of out knowledge concerning the performance of any particular valve substitute originates from one institution; thus, if valve-related complications are more a function of the patient substrate undergoing operation than the prosthesis per se, the usefulness of inter-institutional comparisons would be severely limited. To address this question, the outcome of 2,719 patients after mitral or aortic valve replacement over 12,955 patient-years of follow-up was analyzed by time-dependent multivariate statistical methods with respect to thromboembolic events, anticoagulant-related hemorrhage, valve failure, fatal valve failure, all valve-related morbidity and mortality, necessity for reoperation, and late survival. Many patient-related factors were significant predictors of the probability of certain patient groups for sustaining these valve-related complications. Hence, comparisons of results of valve performance from different institutions may be misleading unless patient populations are comparable.