No research has examined the association between political preferences and residential segregation by educational status. In Turkey, affective polarisation is very high and warrants an examination of whether political preferences are associated with educational residential segregation. This study uses data on Turkey from the 2013 Address-Population Based Registry, the 2011 Census of Population and Housing and voting archives maintained by the Supreme Election Council to examine residential segregation by educational status across the nation's 81 provinces. We find that the segregation between groups at the ends of the educational distribution is the highest. Those with college education are segregated at a moderate level from those with no schooling and a primary-school education. High-school graduates are moderately segregated from those with no schooling. Multivariate analyses reveal that political preferences are significantly associated with educational segregation. The implications of this spatial distancing are discussed for Turkey and other politically polarised societies.