This study inquired into the effect of single-sex schooling on achievement of female adolescents. It used national examination data of 4,787 Polish female students (10 cohorts) attending single-sex and co-educational Catholic schools. We tested two sets of partially contradictory predictions derived from two different theoretical models explaining how and why achievement of female students in the two types of school may differ. The results of cross-classified random-effects regression models showed that after controlling for initial student and school differences female adolescents attending all-girls schools scored higher on the lower-secondary school exam in science and the humanities in comparison to those who attended co-educational schools. However, the difference was statistically significant only for science. The results were fully consistent with neither of the two adopted theoretical models, although provided more support for the one drawing upon peer effects. Although the effect of 17% of the exam scores standard deviation could be considered small, it appeared in the results of a high-stakes exam. Since the examination results were the main criterion for admission to the next-stage school, attending an all-girls school might significantly affect future educational career and job opportunities of young women.