This study established a conceptual framework for capturing the probabilistic nature of travel times with the use of existing traffic simulation models. The framework features three components: scenario manager, traffic simulation models, and trajectory processor. The scenario manager captures exogenous sources of variation in travel times through external scenarios consistent with real-world roadway disruptions. The traffic simulation models then produce individual vehicle trajectories for input scenarios while further introducing randomness that stems from endogenous sources of variation. Finally, the trajectory processor constructs distributions of travel time either for each scenario or for multiple scenarios to allow users to investigate scenario-specific impact on variability in travel times and overall system reliability. Within this framework, the paper discusses methodologies for performing scenario-based reliability analysis that focuses on (a) approaches to obtaining distributions of travel times from scenario-specific outputs and (b) issues and practices associated with designing and generating input scenarios. The proposed scenario-based approach was applied to a real-world network to show detailed procedures, analysis results, and their implications.