A large number of open competitive exams for French State civil service take place on the basis of a two-stage recruitment procedure. The first stage, consisting solely of written tests, preserves the anonymity of candidates, unlike the second stage, which consists almost exclusively of oral tests. The objective of this research is to assess whether this two-stage procedure of competitions guarantees equal treatment between candidates, regardless of their socio-demographic characteristics, and promote a composition of the civil service that reflects the different components of society. Based on a sample of 72 different competitive exams covering almost half the recruitment of officials during 2008-2015, we consider four categorisation criteria: gender, place of residence, marital status and place of birth. The results challenge the capacity of these competitions to treat candidates equally. They reveal inequalities in success to the written tests as well as evaluation biases in the oral tests. However, these evaluation biases tend to partially offset inequalities in success in the written phase, particularly for women and candidates born outside metropolitan France. We also show that the evaluation biases in favour of these two groups have increased in recent years, marked by the rise of diversity policies in the French civil service. Yet, this development does not seem to be linked to a strengthening of the mechanisms for offsetting the inequalities generated by the written tests.