This dissertation presents a new approach for identifying, assessing, mitigating, and managing the risks of failure in complex systems. It describes the paradigm commonly used today to explain such failures and proposes an alternative paradigm that expands the lens for viewing failures to include alternative theories derived from modern theories of physics. Further, it describes the foundation for each paradigm and illustrates how the paradigms may be applied to a particular system failure. Today, system failure commonly is analyzed using a paradigm grounded in classical or Newtonian physics. This branch of science embraces the principles of reductionism, cause and effect, and determinism. Reductionism is used to dissect the system failure into its fundamental elements. The principle of cause and effect links the actions that led to the failure to the consequences that result. Analysts use determinism to establish the linear link from one event to another to form the chain that reveals the path from cause to consequence. As a result, each failure has a single cause and a single consequence. An alternative paradigm, labeled contemporary, incorporates the Newtonian foundation of the classical paradigm, but it does not accept the principles as inviolate. Instead, this contemporary paradigm adopts the principles found in the theories of relativity, quantum mechanics, chaos, and complexity. These theories hold that any analysis of the failure is affected by the frame of reference of the observer. Causes may create non-linear effects and these effects may not be observable directly. In this paradigm, there are assumed to be multiple causes for any system failure. Each cause contributes to the failure to a degree that may not be measurable using techniques of classical physics. The failure itself generates multiple consequences that may be remote in place or time from the site of the failure, and which may affect multiple individuals and organizations. Further, these consequences, are not inevitable, but may be altered by actions taken prior to and responses taken after the occurrence of the failure. The classical and contemporary paradigms are applied using a single embedded case study, the failure of the space shuttle Challenger. Sources, including literature and popular press articles published prior to and after the failure and NASA documents are reviewed to determine the utility of each paradigm. These reviews are supplemented by interviews with individuals involved in the failure and the official investigations that followed. This dissertation demonstrates that a combination of the classical and contemporary paradigms provides a more complete, and more accurate, picture of system failure. This combination links the non-deterministic elements of system failure analysis to the more conventional, deterministic theories. This new framework recognizes that the complete prevention of failure cannot be achieved; instead it makes provisions for preparing for and responding to system failure. / Ph. D.