The prion notion, that a host-encoded protein reacts with itself to become an infectious entity, has become highly favoured as the causal agent in transmissible encephalopathies (TSEs) such as endemic scrapie of sheep, epidemic bovine encephalopathy (BSE), and rare forms of human Creutzfeldt-Jakob Disease (CJD). Much of the claimed "wealth of data" supporting the notion of prions, however, has been irreproducible as well as contradicted by many experimental observations. By 1989, for example, it was already shown that the "infectious prion conformation" could be eliminated without altering either infectivity or agent strain characteristics. Efforts to make the prion protein infectious also continue to fail. All the biological features of TSEs fit better with a conventional viral particle. Furthermore, structural data point to a spherical infectious particle of approximately 25 nm that contains a genome of 1-4 kb in length and probably encodes its own protective nucleocapsid. These viral features can account for the observed host recognition of TSE agents as foreign. Although recognised, TSE agents, like many conventional viruses, can hide in a latent state for many years in lymphoreticular tissues. Previous predictions made by this parsimonious viral interpretation have been accurate, whereas the prion hypothesis has been confounded by several inexplicable realities that have emerged. Recently developed TSE tissue culture models offer a way of identifying rapidly the intrinsic and strain-specific agent molecules so needed for diagnosis and prevention.