In brains of patients with Alzheimer's disease (AD), Aβ peptides accumulate in parenchyma and, almost invariably, also in the vascular walls. Although Aβ aggregation is, by definition, present in AD, its impact is only incompletely understood. It occurs in a stereotypical spatiotemporal distribution within neuronal networks in the course of the disease. This suggests a role for synaptic connections in propagating Aβ pathology, and possibly of axonal transport in an antero- or retrograde way-although, there is also evidence for passive, extracellular diffusion. Striking, in AD, is the conjunction of tau and Aβ pathology. Tau pathology in the cell body of neurons precedes Aβ deposition in their synaptic endings in several circuits such as the entorhino-dentate, cortico-striatal or subiculo-mammillary connections. However, genetic evidence suggests that Aβ accumulation is the first step in AD pathogenesis. To model the complexity and consequences of Aβ aggregation in vivo, various transgenic (tg) rodents have been generated. In rodents tg for the human Aβ precursor protein, focal injections of preformed Aβ aggregates can induce Aβ deposits in the vicinity of the injection site, and over time in more distant regions of the brain. This suggests that Aβ shares with α-synuclein, tau and other proteins the property to misfold and aggregate homotypic molecules. We propose to group those proteins under the term "propagons". Propagons may lack the infectivity of prions. We review findings from neuropathological examinations of human brains in different stages of AD and from studies in rodent models of Aβ aggregation and discuss putative mechanisms underlying the initiation and spread of Aβ pathology.