Somatic mutations are postzygotic mutations which may lead to mosaicism, the presence of cells with genetic differences in an organism. Their role in cancer is well established, but detailed investigation in health and other diseases has only been recently possible. This has been empowered by the improvements of sequencing techniques, including single‐cell sequencing, which can still be error‐prone but is rapidly improving. Mosaicism appears relatively common in the human body, including the normal brain, probably arising in early development, but also potentially during ageing. In this review, we first discuss theoretical considerations and current evidence relevant to somatic mutations in the brain. We present a framework to explain how they may be integrated with current views on neurodegeneration, focusing mainly on sporadic late‐onset neurodegenerative diseases (Parkinson's disease, Alzheimer's disease and amyotrophic lateral sclerosis). We review the relevant studies so far, with the first evidence emerging in Alzheimer's in particular. We also discuss the role of mosaicism in inherited neurodegenerative disorders, particularly somatic instability of tandem repeats. We summarize existing views and data to present a model whereby the time of origin and spatial distribution of relevant somatic mutations, combined with any additional risk factors, may partly determine the development and onset age of sporadic neurodegenerative diseases.