The Fukushima catastrophe tragically epitomizes the limitations of dealing with natural and technical hazards. Remarkably yet; authorities' review of the catastrophe continue to be limited to mistakes and responsibilities of practical risk management. Although state regulations are questioned; technical protection measures verified; and disaster management processes optimized; no deeper discussion about the actual analytical limits of risk analysis has been engaged thus far. What can risk analyses address and what remains beyond their scope? How trustworthy can risk analyses be; and what kind of statements about the future can they actually formulate? This article examines these broader analytical and epistemological boundaries of risk analysis. Drawing on cases of international nuclear risk management; it tests and problematizes how the definition of risk; the methodology of their registration; and the interpretation of results are reaching their limitations in contemporary risk analysis. Following this critical discussion of risk analysis; the article draws conclusions concerning the usefulness and necessity of a differentiated and informed discourse on the potential and the limitations of the risk analysis method; an approach which today enjoys increasing popularity in a variety of policy sectors ranging from critical infrastructure protection to national and international security.