The objective of this study was to determine cut-off points that can be used to differentiate measures of empathy, which would then be classified as high, medium, or low. To do so, we used data from students from 7 medical schools in Colombia, El Salvador, and the Dominican Republic, after determining the psychometric properties of the 3-dimensional model of empathy in the Jefferson Scale of Empathy, S-version (for medical students).This non-experimental descriptive study had a sample that consisted of 6291 students. The structure and factor invariance were analyzed by country and sex. A hierarchical cluster analysis and a bifactorial analysis of variance were applied.The measure of empathy was reliable on the global scale (α = .82; ω = .88). A confirmatory factor analysis showed that the original model was replicable and adjusted to the data (comparative fit index [CFI] = .90; goodness of fit index = .94), while the multigroup analysis allowed to assume an invariant factor structure by country and gender (ΔCFI < .01). Tables were constructed with cut off points for empathy and its dimensions.Our study solves the problem of comparing the scores and the levels of empathy observed in the medical students at different schools of medicine, making said comparisons within and between countries and between genders. The instrument used has adequate psychometric properties and the cut-off values obtained allow the classifying of people with lower or higher levels of empathy.