Résumé : BACKGROUND: In the context of the increasing interest in the study and assessment of emotional abilities in psychology, we translated into French and evaluated the construct validity of the emotional intelligence inventory (EII) designed in English [Psychol Rep; 88 (2001):353-64]. This self-rating scale is modelled along the theoretical constructs of the Mayer, Caruso and Salovey's model [Intelligence; 27 (1999):267-98]; it comprises 41 items tapping the four following factors: (1) empathy; (2) utilization of feelings, (3) handling relationships and; (4) self-control. METHODS: One thousand three hundred and thirty-five students (42.7% men) with a mean age of 19.6 years and affiliated to several school and faculties participated to this Belgian interuniversity study. They were administered the French version of the EII, as well as a series of related questionnaires. Confirmatory factor analyses (CFA) were applied to these data and various fit indices examined in order to assess the factorial adjustment to the data of the four-factor a priori structure. RESULTS: CFA did not support the original 41-item four-factor structure for the scale in French, but a good statistical fit to the data could be obtained with the reduction of the scale to 20 items. The content of the resulting item set, keeping Tapia's four factors, encourages a revision of the domain covered by the subscales. CONCLUSION: Further efforts should be directed at assessing the content validity of the proposed revised scale as a reliable tool in measuring emotional intelligence by self-report.