We decompose annual average years of schooling series for Portugal into different schooling levels series. By estimating a number of vector autoregressions, we provide measures of aggregate and disaggregate economic growth impacts of different education levels. Increasing education at all levels except tertiary has a positive and significant effect on growth. Investment in education does not significantly crowd out physical investment and average years of schooling semi-elasticities have comparable magnitude across primary and secondary levels. (Contains 5 tables.)