 In 2016, then-NRC Chairman Stephen Burns told an audience, Part and parcel of everything that we do is assessment of risk. Public confidence and safety, he asserted, depended on the agency, quote, constantly reassessing how safe is safe enough by practicing what he called the regulatory craft. Burns' observation epitomizes how ingrained risk assessment has become in the NRC's regulatory philosophy since the hard lessons of the 1979 Three-Mile Island accident. The chairman of an accident review commission called the NRC at the time a total disaster, whose almost exclusive focus on preventing large, rare accidents blinded it to the dangers from common mishaps like Three-Mile Island. I'm Tom Wellock, the historian for the NRC. This is the second installment of our history of probabilistic risk assessment, or PRA. In the first video, we covered the development of the NRC's Reactor Safety Study, the world's first PRA. While it was controversial and criticized, the study foreshadowed the chain of failures that led to Three-Mile Island. Since TMI, the NRC has supplemented its traditional safety philosophy with PRA insights to create a hybrid approach called risk-informed regulation. Before TMI, regulators rarely considered accident probabilities. Their approach was called deterministic, meaning they simply determined that a few large-scale, credible events could happen, so-called design basis accidents. Then, they required lines of defense, such as physical barriers and safety systems, to create a defense in depth to the potential escape of radiation, target the big design basis accidents that thinking went, and small ones would be prevented too. At TMI, defense in depth worked. Negligible radiation escaped the containment building, but the reactor was destroyed, not from a big design basis accident, but as the reactor safety study anticipated, from a chain of minor hardware failures, errors, and inadequate control room design, issues that had received only limited regulatory attention before. Post-TMI review committees pressed the NRC to step beyond defense in depth and use PRA to identify the most significant contributors to accident risk. PRA was adept at exploring three key questions, known as the risk triplet. What can go wrong? How likely is it to go wrong? And what are the consequences? Capable of weighing risk with costs and benefits, PRA could broaden NRC capabilities. First, the NRC needed a quantitative safety goal to establish when a reactor was, quote, safe enough for the public. In 1986, the NRC commission adopted the safety goal policy statement for a major accident that set out objectives for immediate deaths and long-term cancers, as well as an objective that's such an accident to occur only once in a million years of reactor operation. Meanwhile, the NRC staff sought to make risk assessment a more comprehensive measure of safety goals. And in 1990, it published New Reg 1150, a state-of-the-art approach to help licensees to build their own individual plant PRAs to improve safety. In the 1980s, the NRC also applied risk insights to unresolve safety issues. For example, the NRC instituted a risk-informed regulation to avert a hypothetical failure of a reactor shutdown system. This occurred after a very real partial shutdown failure at the Browns Ferry Nuclear Power Plant in Alabama. Similarly, the NRC added reliability and probabilistic elements to regulations for station blackouts and auxiliary cooling systems. PRA was tailor-made to address such hardware failures but proved difficult to apply to what are called human factors issues, such as operator error and a poor licensee safety culture. In the 1980s, several licensees displayed a questionable safety culture, including episodes of sleeping on the job by control room operators at the Peach Bottom Nuclear Power Plant. The NRC hoped an objective measure of safety culture could be developed for regulation. But a multimillion-dollar study by Brookhaven National Laboratory concluded safety culture defied quantification. Even today, measuring safety culture remains a challenge for PRA. Other PRA initiatives, however, improved plant operations and management. The NRC was able to allay industry concern that a new regulation on maintenance was too burdensome by encouraging licensees to use risk-informed approaches that also enhanced operational efficiency. In 1995, the NRC Commission recognized the progress and limitations of risk assessment when it issued a new policy statement where it encouraged, quote, the use of PRA in all nuclear regulatory matters. The Commission acknowledged that while PRA could not be the sole basis for safety regulation, it had made enough progress to supplement the NRC's tradition of deterministic safety. Since the turn of the century, risk-informed regulation has supported new initiatives to overhaul the NRC's oversight of operating plants by reducing its subjectivity with comprehensive tracking of performance indicators and aligning enforcement actions to the risk significance of a particular violation. The NRC also applied risk insights to rules for fire protection and in evaluating the importance of safety-related equipment. Risk assessment has also revised our understanding of accident behavior. Research found that the consequences of a major accident was millions of times lower than predicted in earlier studies sponsored by the Atomic Energy Commission and by the NRC. Events such as the 9-11 attacks and the Fukushima accident, however, remind us that plant operators must be ready to cope with unanticipated events that may not be fully captured by a PRA. As a result, defense in depth has evolved too by enhancing the resources available to operators as they cope with the consequences of unexpected events. The history of risk assessment is a story of how our understanding of preventing and coping with major accidents has matured. PRA began as a means to reassure the public that nuclear power was already safe enough. Today, the NRC recognizes that the search for safety is continuous. And with tools like probabilistic risk assessment, the NRC seeks to augment its traditional approach with new insights and information.