Exercise is overwhelmingly beneficial for physical and mental health, but for some people exercise addiction (EA) can develop and negatively impact an individual. This study sought to (a) compare the latent structure of two instruments assessing EA and (b) examine differences in attitudes toward stopping exercise, if required to on medical grounds, among exercise-addicted and non-addicted athletes. In a cross-sectional study, 1,011 athletes competing at different levels completed an anonymous on-line survey. The survey contained Exercise Dependence Scale-Revised (EDS-R), Exercise Addiction Inventory (EAI), and questions on adherence to medical prescriptions to stop exercise. We tested the latent structure of EDS-R and EAI with multigroup confirmatory factor analyses (CFA), across gender and competition level. Finally, we measured the difference of athletes' attitudes toward stopping exercise, if prescribed by a physician. Both instruments showed good fit indexes, even across gender. CFAs on EAI scores showed some violations of measurement invariance across competition level (ΔCFI = .03; ΔRMSEA = .02). On the contrary, CFAs on EDS-R scores did not show invariance violations across competition level (ΔCFI = <.01; ΔRMSEA = <.01). Finally, athletes who reached thresholds for exercise addiction, by means of EDS-R, were more prone to not follow medical prescriptions to cease exercise, independently of the competition level. These results suggest that athletes' answers on the EDS-R seem to be less affected by competition level, compared to EAI. Moreover, EDS-R outcomes could be used to identify individuals who may be unlikely to cease exercise for medical reasons, independently of their competition level. (PsycInfo Database Record (c) 2021 APA, all rights reserved).