Christopher Walsh and colleagues report the identification of large somatic copy number variants (CNVs) in the human brain by single-cell whole-genome sequencing (Cell Rep. doi:10.1016/j.celrep.2014.07.043; 21 August 2014). The authors isolated over 200 neuronal and non-neuronal cells from the post-mortem brains of three normal adults, one fetus with trisomy 18 and one adult with hemimegalencephaly (HMG). These sequences were also compared to the genomes of 24 single cultured lymphoblast cells. Cells from the fetal brain all showed an estimated three copies of chromosome 18, as expected, whereas over 95% of cells from normal brains had no detectable aneuploidy. However, the existence of a small proportion of cells with large CNVs suggests that somatic CNVs may arise normally during development. Although only eight cells from the HMG brain yielded sequences of sufficient quality for analysis, the authors were able to show that only one of the eight cells had a copy number gain at 1q, confirming previous results of mosaic gain of 1q and providing an estimate of its prevalence in diseased brain. The findings highlight the usefulness of single-cell whole-genome sequencing. They also indicate that CNVs in even a minority of brain cells can have large phenotypic effects.