This paper aims to improve the understanding of classroom-based gender differences that may lead to differential opportunities to learn provided to girls and boys in low and high performing primary schools in Kenya. The paper uses an opportunity to learn framework and tests the hypothesis that teaching practices and classroom interactions explain gender gaps in maths achievement in Kenya. The data used is obtained from a cross sectional study involving video recordings of 70 lessons in mathematics, students' scores in a maths test and interviews with subject teachers in Kenyan primary schools randomly selected from six districts. Results show that gender gaps in maths achievement are more evidenced in the area of measurement. The gaps are more pronounced among low achievers in favour of boys. The most revealing finding is that entry achievement level is the main source of gender gaps in maths learning outcomes, implying that girls start at lower levels than boys and this gap is not closed by school. The policy implication to education is that boys have better chances of transition to secondary school and tertiary levels than girls, and consequently, there are broader gender disparities than can be closed by pro-gender education policies.