Research on Alzheimer's disease has been hampered by a lack of good disease models. Mice carrying mutations linked to familial Alzheimer's do not develop the full spectrum of aberrant phenotypes, and neurons from postmortem patients are not readily available. Israel et al. have now shown that patient-derived induced pluripotent stem cells provide an alternative way of modeling Alzheimer's. They found that neurons generated in vitro from fibroblasts of patients with either familial or sporadic disease reproduced important features of the pathology, including increased levels of amyloid-β1–40, phospho-tau and active glycogen synthase kinase-3β (aGSK-3β). To begin to unravel the causal relationships between these three dysregulated proteins, the authors treated the neurons with inhibitors of β- or γ-secretases, which cleave the amyloid precursor protein into several products, including amyloid-β1–40. These experiments revealed that processing of amyloid precursor protein induces phospho-tau and aGSK-3β. Because the neurons generated from one patient with sporadic disease resembled those from the familial-disease patients, the authors suggest that this case of sporadic disease is likely to have a previously unappreciated genetic basis. (Nature 482, 216–20, 2012)