Orthorexia nervosa (ON) is characterised by a pathological fixation on healthy eating. Athletes strive for athletic performance through restricted dietary habits which may lead to the presence of ON symptoms. The aim of the present study was to investigate the underlying structures of two self-report measures of ON, assess differences in ON symptoms between elite and recreational athletes, and determine potential predictors of ON symptomatology. Cross-sectional study of 215 elite (n = 59; mean age 26.71 ± 6.83 years) and recreational athletes (n = 156; mean age 36.13 ± 14.33 years) completed the Teruel Orthorexia Scale (TOS), Eating Habits Questionnaire (EHQ), Short Form Food Frequency Questionnaire (SFFFQ), Eating Attitudes Test-26 (EAT-26), and the Hewitt-Flett Multidimensional Perfectionism Scale Short Form (HF-MPS-SF). Factor analysis revealed two-factor structures for both the TOS and the EHQ. Elite athletes exhibited higher scores for ‘healthy orthorexia’ (HeOr) (P = .016), ‘knowledge of healthy eating’ (P = .038) (EHQ-Knowledge), diet quality score (DQS) (P = .013) and self-orientated perfectionism scores (P = .032) compared to recreational athletes. Multiple linear regression analysis revealed DQS to be a consistent and significant predictor of all ON symptoms (both TOS and EHQ). BMI was negatively associated with HeOr (β = –.21, P < .001). EAT-26 dieting, bulimia and oral subscales predicted ‘orthorexia nervosa’ (OrNe; smallest P = .015) and ‘problems associated with healthy eating’ (EHQ-Problems; smallest P = .012). Other-orientated perfectionism (OOP) was the only perfectionism construct to predict HeOr (β = .23, P < .01), EHQ-Knowledge (β = .34, P < .001), EHQ-Problems (β = .18, P < .01) and EHQ-Total (β = .27, P < .001). Findings suggest TOS is a reliable measure of ON symptomatology in athletes and further refinement of the EHQ is required. Improvements in assessment tools, study methodology and classification of athlete characteristics are required to advance our understanding of ON in athlete populations. None.