Although the role of education in an economy is emphasized in theoretical studies, empirical literature finds mixed results for the relationship between growth and education. We examine the relationship between Gross Domestic Product (GDP) and enrollments in primary, secondary, and high schools, as well as universities in Turkey for 1937-1996, in a multivariate framework. We find that real GDP and educational variables are cointegrated. We also discover different directions of Granger causality between different levels of education and GDP. The generalized forecast error variance decomposition and impulse response analyses seem to confirm the importance of school enrollments in explaining growth. (Contains 3 figures, 5 tables, and 6 notes.)