Recently, there has been increasing interest in comparing educational leadership measures, such as principal school leadership, using International Large-Scale Assessments (ILSAs) data. However, there are doubts about the uniformity of measurement across countries participating in the ILSAs. There are concerns that the robustness and psychometric characteristics of measures are adversely affected by socio-cultural, economic, political, and linguistic diversity across countries. The current study examines the uniformity of cross-cultural model data for the "principal instructional leadership scale" using the framework and data supplied by the Organization for Economic Cooperation and Development (OECD)'s Teaching and Learning International Survey is employed to estimate the conceptual measurement model and test measurement invariance across forty-eight countries. Countries are then divided countries into more homogenous groups, based on their socio-demographic characteristics, to test measurement invariance within these sub-groups. The results of this study reveal that, when testing for the forty-eight countries together, the scale measuring principals' school leadership is invariant across all countries only at an intermediate level (i.e. metric). This means the factor structures and the factor loadings are equivalent across countries, but the item intercepts are not. However, when testing within sub-groups, improvements in cross-cultural comparability are found. This paper concludes by making suggestions on scale improvement, discussing the implications of this study for policymaking and making recommendations for future research.