Introduction: Some authors argue that empathy in medical students undergoes a declining process from the third year of medical school. On the other hand, some authors have found that such a decline does not occur. A working hypothesis is proposed which consists of several coexisting empathic behavior models. Materials and methods. Exploratory and transversal work was included as part of this study. The Empathy scale for medical students (eejm-s) was applied amongst medical students in three medical faculties of Colombia, El Salvador and Argentina between 2018–2019. The mean values of each year of study were subjected to regression tests with previously standardized data. An analysis of variance (Anova) was first performed to determine if the coefficient of the equation was different from zero and sequential Anova was performed to determine the type of curve and distribution of the data. Results: The distribution of the means, through the years of study, presented different types of regression curves between the different universities used in the study, as well as in each of its components for empathy. Conclusion: Differing types of empathic behaviour existed (different regression curves) in the means of the years of study between the different faculties and the working hypothesis proved to be true.