This paper analyses the impact of teacher gender towards students' test results in a blinded Math test administered to students in Catalonia (Spain). The data for this analysis are drawn from a sample of secondary school students who participated in an international blind-test known as the "Mathematical Kangaroo" in 2008. The estimation considers a two-stage procedure since participation on the test leads to the presence of sample selection. Results show a correlation between female teacher gender and student results. Moreover, students with female teachers have a higher probability of participating in the "Kangaroo" test (in this case, the effect being more marked among male students).