OBJECTIVE: Previous epidemiological studies of bulimia nervosa have generated differing estimates of the incidence and prevalence of the disorder. These differences are attributable, in part, to varying definitions of the illness and a range of methodologies. The authors sought to define the prevalence of bulimia nervosa in a nonclinical community sample, examine the clinical significance of DSM-III-R threshold criteria, and examine comorbidity. METHOD: Subjects across Ontario (N = 8,116) were assessed with a structured interview, the World Health Organization Composite International Diagnostic Interview, with specific questions added for bulimia nervosa. Subjects who met DSM- III-R criteria for bulimia nervosa were compared with those who were missing only the frequency criterion (two or more binge-eating episodes per week for 3 months). RESULTS: In this sample, the lifetime prevalence of bulimia nervosa was 1.1% for female subjects and 0.1% for male subjects. The subjects with full- and partial-syndrome bulimia nervosa showed significant vulnerability for mood and anxiety disorders. Lifetime rates of alcohol dependence were high in the full- syndrome group. Rates of parental psychopathologies were high in both bulimic groups but tended to be higher in the subjects with full- syndrome bulimia nervosa. Both bulimic groups were significantly more likely to experience childhood sexual abuse than a normal female comparison group. CONCLUSIONS: This study confirms other prevalence estimates of bulimia nervosa and its comorbid diagnoses from studies that were based on sound methodologies. It also points to the arbitrary aspects of the frequency of binge eating as a diagnostic threshold criterion for the disorder.