Sweden reformed public education in 1991-1993. A system with private school competition was introduced and the production of public education was decentralized from central to local government. One of the aims with this reform was to increase efficiency in the production of education. In this paper, efficiency in Swedish public education is estimated with data envelopment analysis and the efficiency scores are explained by variables reflecting competition and local politics in a second-stage Tobit regression. In 1999-2000 Swedish municipalities could decrease school resources by 8-12% without decreasing outcomes. We find the political majority in the city council to affect efficiency, but we do not find efficiency to be related to private school competition. (Contains 8 tables, 2 figures and 8 notes.)