This article analyses the fact-finding practice of the International Criminal Tribunal for Rwanda (ICTR) to underscore some of the broad challenges faced by the Tribunal and to determine what lessons can be learned from its legacy for the future of complex international criminal trials. It fills a gap in the existing literature by taking a broad assessment of the lessons that can be learned from the ICTR’s fact-finding practice over the course of its lifetime, as the Tribunal adjudges upon its final case. It argues, inter alia, that it is difficult to derive consistent principles on the definition of ‘beyond reasonable doubt’, the requirement of corroboration, and the weight to be given to different types of witness testimony. It also introduces the fundamental concepts of Bayesian probability, and argues that, given that international criminal judgments are inherently probabilistic in nature, the use of Bayes’ Theorem and Bayesian Networks might assist in the decision-making process, in enabling judges to question the strength of their own beliefs as to the truth of a matter. It concludes with some reflections on the function of international criminal tribunals in relation to the historical record of the conflicts upon which they adjudge.