An account of how a rational agent should revise beliefs in the light of new evidence. This theory is set apart from previous belief revision theories by being computationally implementable, provides rigorous mathematical theory of dependency networks and formulates and investigates the complexity of algorithms for rational agents revising beliefs. Rich in helpful and elegant diagrams, accurate index and quite extensive bibliography, the monograph guides the reader through the maze of philosophical issues and logical concepts in a clear and efficient way. At places the narrative unexpectedly changes focus from mathematical definitions to philosophical considerations, from in-depth analysis to broad general perspectives. This feature makes the book a challenging and entertaining read.