This paper is the result of a systematic research review of international comparisons in education by means of International Large-Scale Assessments (ILSA). We ask what research has been carried out and which research results and conclusions are presented in this field of study. We begin by identifying a large set of research publications in the field – more than 11 000 articles are identified by means of search engines for the period 2004-2017. The PISA- and TIMSS-research programmes and the CIVED/ICCS research programme on civics education are chosen – in sum more than 8 000 publications. In the task of assessing research quality, we include peer reviewed scientific articles and primary research on international comparisons. When mapping and synthesizing research it is important to capture arguments and conclusions in a broad field that vary in terms of study- and knowledge objects. A broad result concerns what needs to be explained. We note that the reviewed articles commonly identify achievement gaps over population taxonomies, e.g. classifications in terms of social class or gender, thus pointing to inequities and how they are associated with different kinds of education measures or contextual variations, such as gender inequity coefficients in different countries. To a much lesser extent the differences in efficiency are analyzed, e.g. school performances over educational measures or school systems. We have also found research that analyzes how to redirect or govern students into certain careers – often in science. Examining how these explanations are formulated, the studies often refer to student characteristics, different kinds of education measures and variations in the contextual circumstances. These analyses present what is regarded as significant results based on the strength of association between categories and variables, e.g. how early differentiation in a school system is related to increased social inequity, or how gender gaps differ between national contexts. To our understanding, ILSA research entails a particular kind of statistical analysis and construction of data in order to define the world of education. One conclusion is that the ILSA research field is heterogeneous when the subjects of its research are described. This point is supported by the rather fragmented research communication structure that is captured by analyzing how journal publications quote or reference articles. However, with regard to knowledge objects, there is a homogeneous intellectual organization of ILSA in terms of what can be discussed as style of reasoning. This refers to the ways in which research objects are formulated, how research inquiries are carried out and what are considered as valid statements in this research process. There is an internal relation in the formulation of explanandum and explanans as knowledge objects and accepted procedures for accepting or rejecting statements concerning this relation, e.g. when comparing school performances in different parts of the population. We understand this to be the kind of reasoning that is at work in international large-scale assessments. Such reasoning also sets limits, in that it lends itself to specific analyses and the production of valid statements about the research problem in focus.