OBJECTIVE:To identify risk factors for eating disorders. METHODS:A community cohort study was conducted in Navarra, Spain. A region-wide representative sample of 2862 girls who were 12 to 21 years of age completed the Eating Attitudes Test (40-item version) and other questionnaires in 1997. Girls who scored high in the Eating Attitudes Test-40 were interviewed by a psychiatrist who applied Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria to diagnose prevalent cases of eating disorders. Girls who were free of any eating disorder in 1997 were reassessed after 18 months of follow-up using the same methods. RESULTS:Ninety new cases of eating disorders according to Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria were identified during the follow-up. In the multivariate logistic analysis, a higher risk of incident eating disorder was found for several exposures assessed at the beginning of follow-up, such as younger age, usually eating alone (odds ratio [OR]: 2.9; 95% confidence interval: 1.9-4.6), and frequently reading girls' magazines or listening to radio programs (OR: 2.1; 1.2-3.8 for those most frequently using both media). No independent association was found for television viewing or socioeconomic status. A marital status of parents different from "being married" was associated with a significantly higher risk in the multivariate analysis (OR: 2.0; 1.1-3.5). CONCLUSIONS:Our results support the role of mass media influences and parental marital status in the onset of eating disorders. The habit of eating alone should be considered as a warning sign of eating disorders.