Background: Obsessive-compulsive disorder (OCD) and social anxiety disorder (SAD) frequently co-occur in persons with autism spectrum disorder (ASD). We studied which features distinguish ‘pure’ anxiety disordered patients from those with co-morbid ASD. Method: In a case-controlled design in which groups were matched for age, sex and educational level, patients with OCD or SAD and co-morbid ASD were compared with patients with ‘pure’ (i.e. without ASD) OCD, with ‘pure’ SAD and a control group, using the Autism Questionnaire (AQ), Yale-Brown Obsessive-Compulsive Scales, Liebowitz Social Anxiety Scale, Beck Anxiety Inventory and questions on egodystonia of OC behaviors. Results: No between patient group differences were found on social or general anxiety measures. The AQ subscales communication problems and lack of imagination discriminated best between patients with comorbid ASD and the other groups, ASD patients showing elevated scores, whereas the other patient groups scored equal to controls. On the AQ social skill subscale all patient groups showed elevated scores. On OC symptom severity, pure OCD patients showed highest scores, whereas comorbid ASD subjects scored intermediate between controls and the pure OCD group, the differences being explained by lower obsession severity in the ASD group. There were no differences between the pure OCD and comorbid ASD groups on egodystonia. Conclusion: Patients with comorbid ASD differ from patients with pure OCD and SAD on autism-related problem behaviors, but there is also overlap between groups, possibly reflecting overlapping etiologies. Despite the relatively small sample size, these data strongly suggest that specific autism symptom domains should be assessed to pick up autism-related problems in OCD and SAD patients, and subsequently fine-tune treatment programs for these patients.