Malmö studies in educational sciences nr 75 This thesis explores the standardized assessment of students’ scientific literacy by studying test items, frameworks and result reports from the international comparative study Programme for International Student Assessment, PISA. My research concerns the negative trend observed for Swedish students’ results in science reported in international comparisons since 2000. In this thesis, PISA is considered as a specific kind of practice that acts through a certain rationality, which frames how the measurement is constructed and interpreted. The overall aim is to highlight the epistemological and ontological assumptions that are embedded in the assessment of students’ scientific literacy by PISA. Data was constructed by video documentation of collaborative encounters between 21 groups of 15-year-old students and eleven selected items from the PISA scientific literacy assessment. This method enabled an analysis of the students’ reasoning and the difficulties that arose in these encounters. I also conducted a text analysis of selected frameworks and reports produced under the PISA label, analyzing how science and student performance are discursively constructed in these documents. In this thesis, I examine the similarities and differences between two theoretical approaches: one sociocultural and one sociomaterial. Both are used to explore the embedded assumptions of the PISA scientific literacy assessment. The sociocultural perspective focuses on the students’ situated meaning making as they solve the test questions. The sociomaterial perspective finds inspiration in science and technology studies, and takes a performative stance on scientific practice. This thesis has been formed as a hybrid of a compilation thesis and a monograph. It comprises three articles in English, published or still in the process of publication. The measured knowledge in and about science in PISA are based on onto-epistemological assumptions that are connected to science traditions which are mainly monologistic and representational, whereas this thesis proposes a dialogistic and performative stance. One identified assumption is that language is a neutral transmitter of information, which can be unambiguously communicated and translated without losing or gaining new meanings. Another is the assumption of a single unambiguous, primary frame for interpretation of the test questions, and a third that in PISA, science is assumed to be a socially and culturally neutral object for learning. It appears crucial that the students are able, and motivated, to discern and privilege the scientific perspectives and interpretations while engaging with the complexity of the tasks. My analysis suggests that framing the tasks within fictive, everyday situations, as is significant for PISA, contributes to this complexity. Further, the image of science as portrayed in the test items that were studied, risk reproducing stereotypical images of science and scientifically literate people. To PISA, students are mirrors of the school system and even future society. In the analysis of PISA documents, low performers appeared as threats to future society, due to the risk that they would become ineffective citizens. Meanwhile, other studies assert that standardized comparison is a practice that, when frequently repeated, contributes to lower results and an increasing disillusion of low achievers. It is proposed that PISA, rather than to be seen as a knowledge measurement, should be regarded a knowledge actor.