Chapter 8 compares two (North American) approaches that emerged from the greater social preoccupation with accidents and safety in the 1970s and 1980s. As high-visibility accidents and disasters put the limits of safety engineering and risk management on display, questions arose: Was there a limit to the complexities we could handle? Were there things that we perhaps shouldnot do, or build at all? Normal accident theory (NAT) suggested that some accidents are ‘normal’—and thus in a sense predictable—because they can be traced to the interactive complexity and coupling of the systems we design, build, and operate. Interactive complexity and tight coupling built into the very structure of these systems will generate certain accidents, the theory said, independent of how much risk management we do. Yet there are interactively complex and tightly coupled systems that donot generate accidents, or that havenot yet. So are there characteristics of ‘high reliability’ that can somehow be distilled from the things that these systems are doing? This is what high reliability organizations (HROs), also known as high reliability theory (HRT) suggested. A comparison of the two approaches serves as an introduction to the debate that was triggered by the publication in the early 1990s of The Limits of Safety. Both theoretical schools have had considerable influence on what has happened in safety science since.