Societies are increasingly confronted with High-Impact Low-Probability (HILP) events. These events pose important challenges to societies as they disrupt critical infrastructures (CIs), the backbone of modern societies, leading to cascading and systemic disruptions across interconnected systems. Stress testing has emerged as a prominent approach for assessing system performance under adverse conditions. However, its suitability for addressing characteristics of HILP events, such as uncertainty, urgency, and complexity, remains unclear. This paper presents a scoping review of stress testing methodologies developed to date for terrestrial CIs, with the aim of identifying key methodological elements, with particular attention to the context of HILP events. The review reveals that existing stress testing approaches remain largely sector-specific and domain-specific, rely predominantly on hazard-centric scenarios, and insufficiently account for multi-sectoral interdependencies, dynamic system behavior, and the recovery and adaptation phases of resilience. Moreover, current methodologies tend to emphasize quantitative modeling, involve limited stakeholder participation, and lack mechanisms for iterative learning and adaptation, thereby constraining their relevance in rapidly evolving HILP contexts. In response to these gaps, this study proposes a conceptual framework for stress testing structured around three main stages of pre-assessment, assessment, and treatment. The framework emphasizes cross-sectoral and multi-domain analysis, stakeholder-inclusive and participatory approaches, and explicit consideration of recovery and adaptation processes. This study provides a foundation for advancing stress testing practices that are specifically tailored to HILP events and fosters the resilience of CIs.