Quality of instructional delivery is an important determinant of the extent to which the teaching process has an impact on learning achievement. The literature identifies many contributors of learning achievement including teacher, curricula, teaching, student, home and school environments. On an ‘education production function’ approach, achievement may be modelled as an outcome determined by school and classroom based inputs. The processes of achievement determination are typically assumed, in the absence of detailed information about actual classroom practices. In Kenya, primary schools continue to report differentials in performance on standardized tests even when policy implementers distribute teachers normally to these schools. Such differences in academic performance have been reported even among schools within the same neighbourhood. Schools from the same neighbourhood draw students with similar backgrounds and expose them to teachers who have gone through the same training. This paper hypothesizes that teaching style plays a key role in explaining the differences in academic performance among students between primary schools. Using data from 72 math lessons that were filmed in 72 primary schools in Kenya, the authors demonstrate the extent to which teaching practice explains differences in performance among students and schools. While controlling for individual and school based factors, the paper uses mixed methods to analyse the linkages between teaching practices and learning achievement. The main research finding and policy implication is that students learning achievement can be improved through quality teaching, even when other conditions such as class size are not conducive.