Résumé : The aim of our work was to use the results of the 1st cycle of study in medical school to explore success in 3rd doctorate students in two cohorts (n = 82) of 1st generation students in the numerus clausus system. Because of the homogeneity of mean percentages in the two cohorts and the nil cohort effect on the evolution of performance across study years, the two samples were combined for further analyses. While success (end-of-year percentage) in each of the 3 years of the 1st cycle was positively correlated to success in the other two, the positions (according to ranks) of the students in their group only weakly coincided, if at all, between the 1st cycle and the 3d doctorate. The analysis of correlations between percentages in the 4 years (3 years in the 1st cycle and 3d doctorate) showed that 25% of the variance in the 3d doctorate is shared with outcome in the 1st year and only 7% with that in the 2d year; outcome in the 3d year of the 1st cycle did not contribute at all to outcome in the 3d doctorate. Besides, the chance of an excellent success in the 3d doctorate, defined as being among the 10 best performers, was positively associated to the end-of-year percentage in the 1st year, while neither 2d year nor 3d year results contributed to the prediction, as the logistic regression analysis demonstrated. These results, with a significant contribution of the 1st study year outcome to the success in the 3d doctorate of medical school, were obtained in the context of the numerus clausus and contrast with those obtained in a study preceding this selective procedure, and showing instead the 3d year of the 1st cycle as more predictive. This result could be taken as supporting the very newly established selection process at the end of the 1st study year.