There is very limited evidence globally of the impact that learning in an unfamiliar language has on girls' educational outcomes. In Rwanda, children learn in English medium instruction (EMI) classrooms throughout the basic education cycle, with a recent policy change removing Kinyarwanda as the medium of instruction for the first three years. This paper maps the trends between girls' results in English examinations, their wider educational metrics and the rurality, poverty and gender bias of the district where they live. Results suggest that English results at the primary level cluster around a low pass for boys and girls. However, gender differences become apparent by the secondary 3 English examinations where boys are outperforming girls in every district. Gender differences cannot be explained in our analysis by how rural, poor or gender biased the district is. Our conclusions suggest the importance of going beyond linear assumptions between rurality, poverty, gender bias and girls' performance and we suggest further, in-depth, research of classroom dynamics, English development and out-of-school factors to explore how learning in English is particularly affecting girls.