Normal Accidents

About the book

Normal Accidents examines how the design of high-risk technologies shapes the accidents they produce. Perrow studies nuclear plants, chemical facilities, aircraft, marine transport and other systems, asking why serious failures recur even when operators are competent and safety rules exist. His answer rests on two properties: interactive complexity, where components can interact in unexpected and hidden ways, and tight coupling, where events move quickly and cannot be isolated or reversed.

The argument sorts systems by those properties rather than by industry reputation. Some technologies combine complexity and tight coupling, making certain accidents a structural feature rather than a sign of negligence. Perrow calls these normal accidents. The book then considers what follows for regulation, design and public expectations, including the case for abandoning or radically constraining the riskiest systems.

The book belongs to the safety and organizational-risk literature, and its framework has been taken up widely in discussions of industrial failure. The examples come from an earlier era of technology.

How to read it

Approach the book as an argument built from case material. Track the two central concepts and test them against the examples Perrow supplies. The prose is analytical rather than narrative, and the case discussions are detailed, so readers who want a quick summary will find the middle sections slow.

Pay attention to where the analysis moves from description to prescription. Perrow's claims about which technologies should be curtailed are the most contestable part, and they deserve scrutiny rather than acceptance. Readers familiar with safety engineering can compare his categories with later thinking on risk and resilience.

Who it suits

Managers, engineers, regulators and students of organizational behavior will find a durable vocabulary for thinking about failure. Anyone responsible for safety in complex operations gets a framework for asking whether a system's risks are inherent rather than fixable through better training alone. General readers interested in industrial disasters will also find the case discussions instructive.

Reading considerations

The examples and policy debates reflect the early 1980s, so the book does not address later technologies or regulatory changes. Its distinction between complex and tightly coupled systems is useful but can be applied loosely, and readers should check whether a given case fits both categories. Later editions may include added material, so check which edition you have before relying on the framing.