Of all forms of dementia, Alzheimer's disease is the most prevalent. It is histopathologically characterized by beta-amyloid-containing plaques, tau-containing neurofibrillary tangles, reduced synaptic density and neuronal loss in selected brain areas. For the rare familial forms of Alzheimer's disease, pathogenic mutations have been identified in both the gene encoding the precursor of the Abeta peptide, APP, itself and in the presenilin genes which encode part of the APP-protease complex. For the more frequent sporadic forms of Alzheimer's disease, the pathogenic trigger has not been unambiguously identified. Whether Abeta is again the main cause remains to be heavily discussed. In a related disorder termed frontotemporal dementia, which is characterized by tangles in the absence of beta-amyloid deposition, mutations have been identified in tau which also lead to neurodegeneration and dementia. For Alzheimer's disease the existence of familial forms lead to the proposition of the amyloid cascade hypothesis, which claims that beta-amyloid causes or enhances the tangle pathology. In this review, we describe tau transgenic mouse models in which aspects of the tau-associated pathology, including tangle formation, has been achieved. Moreover, tau transgenic mouse and tissue-culture models were used to test the amyloid cascade hypothesis. In addition, we discuss alternative hypotheses to explain the sporadic forms. The animal and tissue-culture models will provide insight into the underlying biochemical mechanisms of tau aggregation and nerve cell degeneration. These mechanisms may be partially shared between sporadic Alzheimer's disease, the familial forms and frontotemporal dementia. Eventually, Alzheimer's disease may be redefined based on biochemical events rather than phenotype.