This paper undertakes the largest ever quantitative analysis of the distribution of surnames in Great Britain, in order to identify ‘surname regions’. Our empirical analysis uses hierarchical clustering and multidimensional scaling in a computationally intensive analysis of a near complete individual population register of Great Britain. The outcome of this classification is an inductive partitioning of space which, we suggest, reveals a unique and authoritative regional geography that is of use in improving our understanding of population dynamics, as well as of historic and contemporary migration.