In Serbia there are four levels of education: pre-school (until 7 years of age); primary education consisting of eight grades (ages 7-14); secondary education lasting two, three or four years; and higher education lasting from three to six years. Primary education is compulsory for all children. It consists of two cycles, each lasting four years, with the first cycle including grades 1-4, and the second cycle including grades 5-8. About 93% of students who finish primary school continue education in a secondary school. This paper examines the process used in Serbia to select students into different types of secondary schools. In this study the authors are concerned with whether the selection process favours students from higher socio-economic groups or either boys or girls. They linked longitudinal data from Serbian student records to the Programme for International Student Assessment (PISA) data collected on student performance in 2003. Three important findings emerged. First, the distributions of both classroom grades and the selection examination scores are substantially negatively skewed, indicating that the range of scores below the mean extends much further than the high scores extend above it. Second, females and students from higher socio-economic backgrounds do substantially better on both classroom assessments and on the selection examinations. The gap between boys and girls is larger for children from low socio-economic status (SES) backgrounds. Third, after boys enter secondary school, their scores on the PISA mathematics test and their combined scores are much better in relative terms than their earlier classroom grades or selection examination scores. However, this bias was not evident with respect to SES. (Contains 2 figures and 3 tables.)