BACKGROUND AND OBJECTIVE: To validate the best alternative cut-off point of a Spanish version of the Eating Attitudes Test (EAT-40) for the screening of eating disorders cases in the general population. SUBJECTS AND METHOD: The translated questionnaire was administered to a representative sample of 2,734 female students who ranged in age from 13 to 22 years. Participants who scored above 21 points were interviewed, as were a random sample of participants who scored 21 or below. Internal and external validity parameters were estimated for different alternative cut-off scores below the usual EAT score of 30. RESULTS: The best diagnostic prediction was obtained with a cut-off score of 21 points (sensitivity: 73.3%; specificity: 85.1%). A positive predictive value of 20% and a negative predictive value of 98.4% were obtained. CONCLUSIONS: The best diagnostic prediction in a non-clinical setting is obtained with a cut-off score of 21, below that recommended by the authors of the questionnaire.