As a new class of pathogens with unusual properties, prions have been implied in several spongiform encephalopathies mainly affecting farm animals (scrapie, mad-cow disease) and humans (kuru, Creutzfeldt-Jacob disease, fatal familial insomnia) (1). The term prions underlines the feature that the agents would merely consist of a protein (2), perhaps a pathological counterpart of a cellular constituent with poorly defined functions (3-5). In reviewing relevant literature, in order to test the validity of the 'protein-only hypothesis' for coherence and consistency with a molecular paradigm, we have met evident discrepancies, making prion diseases a typical abnormality in a Kuhnian sense. This paper summarizes and analyses the main features of the protein-only model.