The amyloid cascade hypothesis has been the predominant model of molecular pathogenesis in Alzheimer's disease. The finding of tau mutations in other dementias has added weight to the hypothesis as it suggests that tau-pathology is a downstream but essential part of the dementing process. However, some observations remain difficult to reconcile with the hypothesis. In transgenic mice, for example, amyloid generation does not induce the predicted cascade and in man, plaques and tangles are separated temporally and spatially. One alternative possibility is that some common factor, loss of wnt signalling for example, might induce both plaques and tangles.