Using a panel of international student test scores 1980–2000 (PISA and TIMSS), panel fixed effects estimates suggest that government spending decentralization is conducive to student performance. The effect does not appear to be mediated through levels of educational spending.