Objective. To determine the association between the non-use of health services due to mistreatment based on ethnicity in Peru. Methods. Secondary analysis of National Household Survey (ENAHO) for 2019. We used Poisson generalized linear models for complex samples, and calculated crude and adjusted prevalence ratios (aPR) with their 95% confidence intervals (CIs). Collinearity relationships between variables were evaluated in the adjusted model. Results. We analyzed 23,242 subjects. The prevalence of mistreatment was 1.4%, 95% CI 1.2–1.6. Belonging to Quechua or Aymara ethnicity was associated with higher prevalence of mistreatment (aPR 2.60, 95% CI 1.75–3.87 and aPR 2.98 95% CI 1.68– 5.27, respectively). Further, the same was found if patient does or does not speak a native language (aPR 2.28, 95% CI 1.66–3.14). Conclusions. Being a Quechua patient or speaking a native language increased by two times and being of Aymara ethnicity increased by three times the probability of not going to health services due to mistreatment. The government should be encouraging cultural competency training for staff and considering staff more representative of the population.