Much of the controversy surrounding school vouchers, and privatization schemes generally, stems from concerns about social stratification. This paper identifies the form and magnitude of selection effects in a means-tested New York City voucher program. It compares students who applied for vouchers, with the eligible population of public-school students; those who initially used vouchers, with those who declined them; and those who remained in private schools, with those who eventually returned to public schools. Differences along the lines of ethnicity, residential mobility, mother's education, and income are observed. In addition, specific aspects of a child's education--parental satisfaction, school uniform requirements, and larger class sizes--all increased the length of time voucher students remained in private schools. Throughout the program's life span, however, the largest and most consistent effects revolved around families' religious identity and practices.