The aim of this study is to investigate how school choice opportunities affect school segregation. Theoretically, free school choice may affect school segregation in different directions, depending on its design, baseline residential segregation, and underlying preferences for separation. Our test case is the Swedish voucher-based free school choice system, and we utilize Swedish population register data that include 13 cohorts of ninth graders, with geocoded information on schools, their socioeconomic and ethnic composition, and the composition of the neighborhoods surrounding the schools. To identify causal effects of school choice opportunities, we treat fixed school areas as the unit of analysis, and we follow these over time to net out time-invariant area-level confounders. Within-area panel estimates indicate segregation based on both ethnic and educational background to be strongly affected by school choice opportunities. Increased choice opportunities lead to increased school segregation, to a large extent because of a higher propensity among native children and children with well-educated parents to attend newly established (non-public) independent schools. The segregating impact of school choice opportunities is uniform across school areas with different socioeconomic and ethnic profiles, but school segregation increases much more in residentially integrated areas as a consequence of an increase in school choice options. The lesson to be learned from the Swedish case is that large scale school voucher systems need to be designed to include mechanisms that counteract the strong segregating forces that such systems appear to produce.