That discrimination cases are particularly difficult to win in court is a common finding of legal scholarship in many countries. The problem of proof and a tendency of judges to restrictively interpret the law are often put forward to explain this situation. Yet, such observations are very general and leave unresolved the question whether, among legal actions for discrimination, certain categories of actions are more likely to fail or succeed than others, and if so why. In the US, the seminal socio-legal research undertaken by Berrey, Nelson and Nielsen has shed important light on the variety of factors susceptible of influencing the outcome of discrimination cases. In Europe, by contrast, similar empirical data are virtually absent. The research presented in this paper aims to fill this gap. Taking Belgium as a case-study, we seek to assess through a statistical analysis the correlation between a set of variables and the outcome of discrimination litigation in the field of employment. Two categories of factors are taken into account: (1) factors relating to the case itself (ground at stake; type of discrimination alleged; nature of the employment dispute); (2) factors relating to the parties (type of plaintiff and type of defendant). Given the absence of a centralized case law data base in Belgium, we created for the first time in Belgium an original data base which includes all judgments we could identify relating to discrimination based on the six prohibited grounds under European Union law (sex, race or ethnic origin, disability, religion/belief, sexual orientation and age) decided by employment courts between 2009 and 2019. The judgments – 596 in total – were collected from tribunals themselves as well as other sources (equality bodies, legal journals and practicing lawyers). Our research shows that each of the variables studied has a significant impact on the outcome of the case. Concerning first the characteristics of the case, we unveil a striking disparity in the success rate depending on the ground ...