To establish diagnostic validity and usefulness of EAT-26 for the risk assessment of Eating Disorder (ED) in a female population.Observational validation study questionnaire.Performed in a Medellin city community care level of mixed (public and private) psychiatric consultation.Twenty five subjects aged 15 to 25 with DSM-IV-TR diagnostic criteria for anorexia and bulimia nervosa and 111 controls without ED.The case sample was for convenience and in controls an aleatory simple one. Gold standard (structured psychiatrist interview confirming the fulfillment of ED case inclusion criteria) was compared with EAT-26 questionnaire; reliability was assessed, cultural, semantics and factorial validation was made and the best cut-off score was established with the ROC curve.Four domains remain in the instrument: bulimia, dieting, food preoccupation and oral control. The Cronbach's alpha was 92.1% and a score of 11 and over is the best cut-off (sensitivity 100%, and specificity 85.6%).This modified and abbreviated EAT-26 questionnaire is an ideal multidimensional instrument for ED screening in risk population, with excellent reliability and sensitivity values and satisfactory specificity. EAT-26 is a useful measure to be considered when strategies for ED early detection are implemented in young women.