Recent achievement test results show that Turkish students have been performing poorly compared to students from other countries. Using science literacy results from the PISA 2006 survey, we aim to measure the determinants of student achievement in Turkey within the education production function framework. We find that program types have large effects on student achievement. Since many students are allocated across program types as a result of a centralized exam, we try to control for past academic achievement and also for differences in the average socioeconomic background of students in a given school. The results indicate that students cluster in schools according to their socioeconomic background, and previous academic record is important, however, the effects of program types persist. Policies to augment socioeconomic background, such as conditional cash transfers and lifelong learning programs that prioritize parents will increase educational output. (Contains 4 tables.)