Despite the large number of studies that draw on Programme for International Student Assessment (PISA) microdata in their analyses of the determinants of educational outcomes, no more than a few consider the relevance of geographical location. This paper examines the differences in educational outcomes between students attending schools in rural areas and those enrolled in urban schools. We use microdata from the 2006, 2009 and 2012 PISA survey waves for Colombia. The Colombian case is particularly interesting in this regard due to the structural changes suffered by the country in recent years, both in terms of its political stability and of the educational reform measures introduced. Our descriptive analysis of the data shows that the educational outcomes of rural students are worse than those of urban students. In order to identify the factors underpinning this differential, we use the Oaxaca-Blinder decomposition and then exploit the time variation in the data using the methodology proposed by Juhn-Murphy-Pierce. Our results show that most of the differential is attributable to family characteristics as opposed to those of the school. From a policy perspective, our evidence supports the need to complement measures of positive discrimination of rural schools with actions addressed at improving household conditions.