This article examines the effects of education system types and characteristics on changes in the distributions of literacy and numeracy skills during the upper secondary phase of education and training. Whereas there is a substantial literature on system effects on skills during the primary and lower secondary phases of education, much less has been written about these effects in relation to the upper secondary phase. This article reports on research using quasi-cohort data for 15-year-olds in PISA (Programme for International Student Assessment) and 18- to 20-year-olds in the Survey of Adult Skills to test the effects of a range of system indicators on changes in skills distributions during the upper secondary phase. Consistent with some dominant theories, our difference-in-difference analysis identifies a range of system characteristics associated with skills inequality reduction in upper secondary education and training, which relate to 'system standardisation' and 'parity of esteem', and which can explain why some 'system types' are more effective than others in reducing skills inequality.