Gender gaps in school enrollment and achievement are usually investigated as separate facets of gender inequality in education. This study suggests that they might be linked by design. If boys and girls differ in their school enrollment rates, the student population will be gender biased. Since out-of-school adolescents tend to be less advantaged than in-school ones, school-based large-scale assessments effectively compare a more fully represented group of one gender with a less represented and more advantaged group of the other one. This should shift student gender achievement gaps so that they favor the latter group. In country-level regression models using data from all PISA cycles, we indeed found evidence of a small, negative association between gender gaps in secondary school enrollment and gender gaps in student achievement. This finding is robust across different achievement domains and specifications. We discuss the study's limitations and implications for applied research with PISA data.