This article introduces a new maximum likelihood estimator for analyzing durations in continuous time war of attrition games, where players compete over a prize with incomplete information about each other's strength.

Key Finding: The model shows strong players persist while weak ones exit via an increasing hazard rate until an endogenous point.

Duration Modeling Approach: We account for two unobserved durations: one ending abruptly and another continuing indefinitely between strong players.

Real-World Application: Using this estimator to analyze Senate filibuster lengths and international crisis durations reveals insights into strategic persistence in political contexts.