The immune system may be viewed as a network of interacting cells producing cooperative behavior with the properties of memory and learning. Like the central nervous system, it can be studied as a parallel computing network, endowed with cognitive capabilities. The contributions in this book discuss the various formal approaches available today to analyze the behavior of immune networks. The first part of the book presents empirical data on some molecular and cellular interactions, mainly idiotypic-antiidiotypic, which constitute the substrate for the connections between different cell populations. Automata network behavior can be used to model not only idiotypic-antiidiotypic interactions but also other interactions between different classes of lymphocytes as well as receptor-antigen interactions. Thus the concept of immune networks receives a more precise meaning. Two different approaches are then discussed. Part two discusses the "large" network approach made of all possible idiotypic-antiidiotypic interactions with cross-reactivity. Part three discusses the small network approach, better fitted to represent and analyze specific data on limited interactions. The book presents the first applications of neural network computation to immunology.