Resumen: Objective: To verify that the factorial structure of the data observed in doctors corres-ponds with the theoretical factorial structure of empathy. Estimate and compare the levels of empathy in the evaluated medical specialties. Materials and Methods: The study was carried out in the canton of Cuenca (Ecuador) in 2019. It has a descriptive and cross-sectional design. The sample consisted of 223 physicians who worked in the Public Sector, who voluntarily answered the Jefferson Medical Empathy Scale. Dependent variables: levels of empathy and their dimensions. Independent: gender, level of care, and specialty. Normality and homoscedasticity, Confirmatory Factor Analysis, Invariance between groups, trifactorial and one-factor analysis of variance, Stu-dent’s t and Cohen’s d were estimated. The level of significance was α≤.05. Results: An adequate internal consistency is observed, the three-dimensional model of the original instrument is confirmed. There are no differences between genders (p>.05), but there are between types of care and between specialties (p <.05). Conclusion: The observed data have the theoretical factorial structure associated with the instrument used to measure empathy. The multigroup analysis makes it possible to assume factorial invariance of empathy between the genders without being able to specify a model according to types of patient care. The levels of global empathy were higher in primary care compared to hospital care and higher in the specialties of Pediatrics and Family Medicine (united).