OBJECTIVE To evaluate the validity of two definitions of partial eating disorders, the Eating Attitudes Test (EAT) and a combination of the EAT and relevant criteria of the DSM-III-R, and to examine their association with factors related to anorexia nervosa and bulimia nervosa. METHODS Questionnaires on eating behaviors, depression, obsessionality, and impulsivity were distributed to 534 female high school students. Demographic, psychosocial, and physical parameters and the subjects' height and weight were also recorded. RESULTS EAT scores revealed maladaptive behaviors in 18% of the subjects: the combination of the EAT and DSM-III-R criteria identified 20.8% of the subjects as having partial anorexia nervosa and 11.3% as having partial bulimia nervosa. Both definitions were significantly associated with risk factors for clinical eating disorders: high weight, weight fluctuations, dieting, menstrual disturbances, high level of depression and obsessionality, and preoccupation with eating in the family. Partial bulimics fared worse on most of these parameters. Partial anorectics were not more psychologically distressed than normal subjects. CONCLUSIONS The validity of both definitions of partial eating disorders is supported by their similar and significant associations with known risk factors for the development of the clinical syndromes. Partial bulimics are similar to patients with bulimia nervosa in the level of many eating-related disturbances and in depression, obsessionality, and impulsivity. Partial anorectics, like anorectic patients in clinical settings, tend to minimize their problems. The relevance of partial eating disorders to the later development of the full-blown clinical entity is still not established.