The OECD's Programme for International Student Assessment (PISA) is increasingly depended upon by education policy makers to provide reliable measures of their country's education system against international benchmarks. PISA attempts to provide efficient, scientific and technical means to develop educational policies which achieve optimal outcomes (Berg & Timmermans, 2000, p. 31). This kind of scientific evidence is seen by policy makers as being free of prejudice and ideology. Science is expected to represent the truth, state universal facts and make predictions. (...) Thus PISA seeks to rank countries' performances, work out future scenarios and offer policy direction. By what means does PISA gain knowledge and speak with confidence about diverse cultures and distant nations? How does it acquire a ‘voice from nowhere’ (Haraway, 1988; Suchman, 2000), and become a modern-day Oracle that countries might consult for policy advice? Modelled on early actor-network accounts of laboratory life, this ethnography traces how PISA knowledge comes to be made, guided by interview data with two ‘insiders’ in the ‘PISA laboratory’. It traces the translations and the circulating reference that turn PISA into a ‘centre of calculation’. It highlights how human and non-human entities are imbricated in the assembling of scientific facts and argues for a suspension of the divide between ‘science’ and ‘politics’. In the process, the paper offers an empirical instantiation of how some concepts from actor-network theory may be applied in the field of education policy, and ponders the implications of such an understanding for evidence based policy making. (shrink)