The central theme in prion diseases is the conformational transition of a cellular protein from a physiologic to a pathologic (so-called scrapie) state. Currently, two alternative models exist for the mechanism of this autocatalytic process; in the template assistance model the prion is assumed to be a monomer of the scrapie conformer, whereas in the nucleated polymerization model it is thought to be an amyloid rod. A recent variation on the latter assumes disulfide reshuffling as the mechanism of polymerization. The existence of stable dimers, let alone their mechanistic role, is not taken into account in either of these models. In this paper we review evidence supporting that the dimerization of either the normal or the scrapie state, or both, has a decisive role in prion replication. The contribution of redox changes, i.e., the temporary opening and possible rearrangement of the intramolecular disulfide bridge is also considered. We present a model including these features largely ignored so far and show that it adheres satisfactorily to the observed phenomenology of prion replication.