The purposes of this study were to examine the percentage of female elite athletes and controls with disordered eating (DE) behavior and clinical eating disorders (EDs), to investigate what characterize the athletes with EDs, and to evaluate whether a proposed method of screening for EDs in elite athletes does not falsely classify sport-specific behaviors as indicators of EDs. All athletes representing the national teams at the junior or senior level, aged 13-39 years (n=938), and age-group matched, randomly selected population-based controls (n=900) were invited to participate. From the screening data, a random sample of athletes (n=186) and controls (n=145) were subjects for a clinical interview. More athletes in leanness sports (46.7%) had clinical EDs than athletes in non-leanness sports (19.8%) and controls (21.4%) (P<0.001). Variables predicting clinical EDs, and thus candidates for valid screening procedures, were menstrual dysfunction in leanness athletes, self-reported EDs in non-leanness athletes, and self-reported use of pathogenic weight control methods in controls. Hence, statistically based risk factors are not universally valid, but specifically related to athletes and non-athletes, respectively.