International large-scale assessments are very successful. One key factor of this success is their rigorous methodological and psychometric basis. Because education systems worldwide are subject to rapid changes, international large-scale assessments need to evolve as well. We describe five current methodological challenges that should be addressed so that large-scale assessments can continue to provide highly useful information on educational outcomes in the future. First, new or changed constructs should be adopted, and constructs with declining importance should be dropped from the assessments. Second, the heterogeneity of student performance within and between countries should be better accounted for. This can be achieved by completing the introduction of computerized adaptive testing into international large-scale assessments and making full use of computers to optimise the testing and scaling process. Third, more analytical effort should be invested in the measurement and modelling of context variables, mainly by applying latent variable.