The objectives of this study were to develop a French version of the Pathological Narcissism Inventory (PNI ; Pincus et al., 2009) and to examine its internal structure. This French version was developed using a forward-backward procedure. Two two-person teams worked independently, one translating from English to French, the other translating back from French to English afterwards. Next, the two teams were assembled, and a new person joined them, and this committee developed the final French items on the basis of this work. The translated measure was then administrated to French speaking participants and the internal structure of this version was examined through a series of Factor Analyses. First, Confirmatory Factor Analysis (CFA) failed to reproduce the original one-level structure of the English version, as well as the two-level structure identified by Wright et al. (2012). Principal Component Analysis (PCA) was then performed to identify the internal structure of our data, which appears very similar to the English one and two-level structures. The sample was then randomly divided in two and the best fitting model was compared with PCA and CFA in the two subsamples. Finally, a multilevel model with fewer items best fits the data.