Exercise addiction is widely studied, but an official clinical diagnosis does not exist for this behavioral addiction. Earlier research using various screening instruments examined the absolute scale values while investigating the disorder. The Exercise Addiction Inventory-3 (EAI-3) was recently developed with two subscales, one denoting health-relevant exercise and the other addictive tendencies. The latter has different cutoff values for leisure exercisers and elite athletes. Therefore, the present 15-country study (n = 3,760) used the EAI-3 to classify the risk of exercise addiction (REA), but only if the participant reported having had a negative exercise-related experience. Based on this classification, the prevalence of REA was 9.5% in the sample. No sex differences, and few cross-national differences were found. However, collectivist countries reported greater REA in various exercise contexts than individualist countries. Moreover, the REA among athletes was (i) twice as high as leisure exercisers, (ii) higher in organized than self-planned exercises, irrespective of athletic status, and (iii) higher among those who exercised for skill/mastery reasons than for health and social reasons, again irrespective of athletic status. Eating disorders were more frequent among REA-affected individuals than in the rest of the sample. These results do not align with recent theoretical arguments claiming that exercise addiction is unlikely to be fostered in organized sports. The present study questions the current research framework for understanding exercise addiction and offers a new alternative to segregate self-harming exercise from passionate overindulgence in athletic life.