Researchers have long been investigating the gender difference in achievement in mathematics. Among different school characteristics, the impact of a school’s gender composition on students’ mathematics achievement is still controversial. Building on the gender stratification hypothesis, this study, using data from the OECD’s 2006 Programme for International Student Assessment (PISA), explores the role of single-sex schooling for mathematics test scores in sixteen countries. Similar to earlier research, the current study found that single-sex schooling was not beneficial to student performance in mathematics, for all selected countries. The role of single-sex schooling may be complicated with respect to specific contexts in different countries and different stages of social and economic development, including gender stratification at the societal level, a competitive and pressured environment in selective single-sex schools, orientations of single-sex schools, and the curriculum and pedagogy. However, the size of the single-sex school sector in a country, as Baker, Riordan and Schaub (1995) argued, did not have an association with the effect of single-sex schooling. In particular, we discuss the gendered curriculum of Hong Kong schools that constrains girls’ achievement in mathematics and counteracts the advantage of selectivity at girls’ schools.