The PISA study has revealed a gender gap in the mathematical performance of Spanish students. The aim of this paper is to determine the different impact of variables related to student characteristics and the schools they attend on mathematics performance by gender. We consider the Spanish sample of the 2018 edition of PISA which, by presenting a hierarchical structure, allows us to apply a methodology based on multilevel regression at two levels. The results obtained indicate that speaking a language other than Spanish or any other co-official language at home and the decrease in the student-teacher ratio have a negative influence, exclusively, on the mathematics performance of boys, while immigrant status and public ownership of the school affect, also negatively, only the mathematics performance of girls. On the basis of these results, gender-specific measures could be proposed to mitigate the effect of the variables that undermine the mathematical performance of students in Spain.