The northern provinces of Burundi have suffered from an inferior education system since independence. This paper shows that the current, northern-led regime has chosen a drastic way to reverse that subordination. The national test (Concours National) at the end of primary school is at the heart of the matter. Using the universe of individual test score data which can be used to construct a school-level panel and applying differencein-differences analysis, the paper shows strong improvements in test scores in northern versus southern schools since the ruling party won an absolute majority in the 2010 elections. Immediately after these elections, schools situated in very poor, rural areas in the north scored as high as schools in non-poor areas of the capital. The paper finds that increased success rates, improved mean test scores and decreased standard deviations are explained by the percent of votes at the municipality level obtained by the ruling party in the 2010 parliamentary elections. Controlling for school budget and cohort size variables does not change the results. The latter are interpreted in the political economy of education reform in Burundi and considered as a case of ethno-regional favoritism in Africa.