This paper examines the impact of wage dispersion on firm performance, measured by value-added per worker, in 2,354 Belgian firms of at least 25 workers. Using a matched employer-employee unbalanced panel data from 2002 to 2010, a conditional indicator of intra-firm wage dispersion is computed following the Winter-Ebmer and Zweimüller’s (1999) methodology. The results support the existence of a positive and hump-shaped relationship between wage dispersion and firm performance. The potential issue of simultaneity between the productivity and conditional dispersion is addressed to limit the endogeneity threat. Furthermore, the analysis shows that increasing conditional wage dispersion may raise value-added per worker in Belgium, particularly among white-collar workers. This may be explained by higher monitoring costs and higher production - effort elasticity for these workers who generally perform more complex tasks. Lastly, estimating the relationship for each of the three Belgian regions shows interesting differences, although insignificants, in the relationship between conditional wage dispersion and productivity across the regions. Compared to other regions, a stronger positive relationship is observed in Brussels in part, due to a more diverse and concentrated workforce, which further lands in higher competition between workers. Additionally, we find that while higher wage dispersion primarily leads to higher productivity, at diminishing returns, in Flanders, in Wallonia, worker productivity starts declining sharply after a certain level of conditional wage dispersion. These observations tend to support the argument that different ideologies and values, observed via political preferences, may influence the workers’ behaviour at different levels of wage dispersion. ; Master [120] en sciences économiques, orientation générale, Université catholique de Louvain, 2017