This study aims to evaluate and compare the scores of a self-reported measurement of orthorexia nervosa among Turkish people who engage in different levels of physical activity (sedentary individuals, recreational athletes and competitive athletes). Data for this study were collected through a self-administered online questionnaire between March and April 2020 in Turkey. The questionnaire consisted of two parts: personal information (sociodemographic and anthropometric information, training and dieting regime) and the ORTO-11 Scale. Only the items in the revised scale (ORTO-R) were used for analysis. Measurement invariance was tested and latent means were compared across different activity groups, sexes and levels of education. Analysis was done with a total of 877 participants (514 sedentary, 271 recreationally active, and 91 competitive athletes). Mean age of participants was 29.12 ± 11.15. Sedentary people were less likely to follow a particular diet. Configural, metric and partial scalar invariance were proven across three groups of physical activity and the difference between latent means was found nonsignificant (competitive vs sedentary: CR = - 0.53, p = 0.592; competitive vs recreational: CR = 0.25, p = 0.724; sedentary vs. recreational: CR = - 1.35, p = 0.183). Configural, metric and full scalar invariance were proven across sexes and different levels of education, and difference between latent means was found nonsignificant (CR = - 1,11, p = 0.272; CR = - 0.53, p = 0.587, respectively). In conclusion, people who compete at sports events, exercise regularly at a recreational level, or lead a sedentary life have similar scores in ORTO-R. Also, females did not score higher than males. Level III, case-control analytic study.