Comorbid personality disorders in eating disordered patients may seriously affect the treatment and course of their illness. Several studies show such a comorbidity, though with inconsistent findings. Qualitative reviews attribute this to methodological shortcomings, but the qualitative method may itself create new shortcomings. To circumvent this, the present, more extensive review applies a meta-analytic approach. Using the databases MEDLINE and PSYCHLIT, the 28 articles published between 1983 and 1998 that presented empirical evidence for an eating disorder and personality disorder comorbidity suitable for meta-analysis were included. We found a higher proportion of eating disordered patients with any personality disorder (average proportion = 0.58) related to comparison groups (average proportion = 0.28). Compared with anorexia nervosa patients, a higher proportion of patients with bulimia nervosa had a concurrent cluster B personality (average proportion = 0.44) and a borderline personality disorder (average proportion = 0.31). However, no differences between anorexia nervosa and bulimia nervosa patients in proportions of cluster C were found (average proportion = 0.45 and 0.44 respectively). Patients with eating disorders and patients with bulimia nervosa in particular, should be routinely assessed for a concurrent personality disorder using structured clinical interviews. In future research, more stringent assessment procedures are highly recommended to address the question of causality between eating disorders and personality disorders, and how eating disorder symptoms and personality disorder symptoms are related to treatment effects.