[en]This paper aims to study the determinants of corruption by examining specificities relating to the region and the level of economic development. Starting from a cross-sectional study on 130 countries, we rely on the Bayesian Model Averaging (BMA) approach to address the issue of model uncertainty and identify the key determinants of corruption according to the level of development and the region. Our results highlight the need for specific remedies in the fight against corruption given the regional, sociocultural, economic and institutional specificities. Indeed, the key determinants of corruption in sub-Saharan Africa are not the most relevant in the East Asia and Pacific region. Similarly, the most important determinants in developed countries are not the most worrying in developing countries.[/en]