Like any simplified representation of the real world or prediction of the future, the mortality assumptions used in the life insurance industry are inherently imperfect, and with enough time, all will prove to be wrong in one way or another. But the challenge they present remains, and new tools and techniques provide opportunities for improvement. This session will discuss approaches, lessons, and findings from two types of predictive models applicable to life insurance mortality and compare their use to traditional methods, including discussions of challenges in implementation and interpretation.