It is one of the pioneer books in the field: the first edition was published in 1981. In this book (and his multiple other works), Peter J. Huber created a foundation for the further development of robust methods.

While I definitely recommend this book to everyone who wants to become an expert in robust statistics, I do not recommend it as the first book on this topic. The book is quite advanced and written for people with a strong mathematical background. For example, the first topic that is discussed just right after the introduction chapter is “The Weak Topology and its Metrization” (Levy, Prohorov, and the bounded Lipschitz metrics; Frechet and Gateaux derivatives; Hampel’s Theorem). This is an interesting and peculiar way to start discussing robust statistics. Most of the terms and notation symbols are assumed to be known by the reader, so the author doesn’t disturb you by providing definitions and explanations. The second edition has multiple improvements, but the book structure and style are the same.

Conclusion: the book is great but only for advanced readers; not for beginners.