This study aims at investigating how efficient are Brazilian States, including their municipalities, and Federal District, in employing resources in primary and secondary education.The methodology employed over a fourteen years — 1995/2008 — balanced panel data, the so-called Three-Stage Approach, goes beyond simple evaluation based on minimizing inputs and maximizing outputs by incorporating random effects and exogenous constraints that may affect performance of decision-making units into evaluation based on a mixed technique of data envelopment analysis – DEA and stochastic frontier analysis – SFA. The framework laid out by last stage results was significantly different from those found at conventional DEA.