The network theory was proposed 21 years ago, attracting then much interest as applied to the regulation of (clonal) immune responses. The first 10 years of 'idiotypic network' research have thus addressed questions that were already appropriately solved by the clonal selection theory, leading to a justifiable loss of its impact. In contrast, 'second generation networks', concentrate on systemic properties that emerge from a network organization, thus providing a framework for several major questions that seem to supersede clonal solutions: the developmental 'learning' of self antigenic composition and the maintenance of the respective 'memory' (self-tolerance), repertoire selection and the homeostatic regulation of lymphocyte numbers, natural immune activities that are independent of external antigens, the physiology of autoreactivity. The immune network may well contribute solutions to autoimmune diseases, where clonal approaches, classical and modern alike, have failed.