In this paper we study the connection between growth and human capital for the Italian regions in a convergence regression framework. We confirm the usual result on Italian regional convergence: using a sample 1963-94, we show that this process began to diminish or fail after about 1975. We include a measure of human capital in the convergence regression as a stock rather than a flow. We find this variable is significant if and only if we control for the size of the Public Sector. Moreover, we find educating women leads to faster growth. The Public Sector is itself strongly negative. We also find exhaustive evidence of two convergence clubs: the South and the North-Centre. Both areas seem to be converging to different levels of GDP per capita and the role of the human capital is different in the two "clubs". While in the North-Centre human capital is never positive and significant our empirical analysis shows large returns of human capital in the southern area of the country. Even if this results do not confirm the usual assumptions of most of the theoretical models on growth and education they are consistent with the microeconometric literature on private returns to schooling, where returns to education are higher the lower the level of education and in more disadvantaged areas.