The paper reports on a study aimed at investigating perceived gender differences in performance in science at secondary school level, as well as beliefs on possible underlying causes for these differences. The study is situated within the interpretivist paradigm and uses a typology of factors drawn from the Educational Effectiveness Research model as a conceptual framework. Six purposively selected secondary schools participated in the study. Focus group discussions were conducted with 48 learners and 22 teachers, and a questionnaire was administered to the six principals of the participating schools. The study indicates that learners and teachers have mixed views on which gender performs best in the sciences. All principals perceived boys as performing best. The perceived reasons for gender differences in performance are wide ranging and reflect broadly student and classroom factors. They include learners' self-efficacy including attitudes towards science; gendered thinking and aptitude; diligence and perseverance; home experiences and culture; language proficiency; socio-economic challenges and the use of discussion as a learning strategy. The reasons for gender difference in performance are perceived to impact the achievement of girls more negatively than boys.