Imagine a robot trying to size up a difficult situation, to find a way of responding. Its sensors receive streams of information from which it tries to reach judgements. If it relies on deduction alone, it will not get far, no matter how fast its inference engines; for even the most massive information is still typically incomplete: there are relevant issues that it does not resolve one way or the other.

The robot, or human agent for that matter, needs to go beyond these limits. It needs to `go supraclassical', inferring more than is authorised by classical logic alone. But such inferences are inherently uncertain. They are also nonmonotonic, in the sense that the acquisition of further information, even when consistent with the existing stock, may lead us to abondon as well as add conclusions.

Nonmonotonic logic is the study of such reasoning and has been the subject of intensive research for more than two decades. But for the newcomer it is still a disconcerting affair, lacking unity with many systems going in different directions.

The purpose of this book is to take the mystery out of the subject, giving a clear overall picture of what is going on. It makes the essential ideas and main approaches to nonmonotonic logic accessible, and meaningful, to anyone with a few basic tools of discrete mathematics and a minimal background in classical propositional logic. It is written as a textbook, with detailed explanations, examples, comments, exercises and answers. Students and instructors alike will find it an invaluable guide.