Self-report measures are often used in research and clinical practise as they efficiently gather a large amount of information. With growing numbers of self-report measures available to target single constructs, it is important to revisit one's choice of instrument to be sure that the most valid and reliable measure is employed. The Maudsley Obsessive–Compulsive Inventory (MOCI) and the Obsessive–Compulsive Inventory-Revised (OCI-R) were administered to 223 female participants: 30 inpatients with anorexia nervosa (AN), 62 community cases with AN, 69 community cases weight restored from AN and 62 healthy controls. Both measures distinguished between clinical and healthy groups; however, the OCI-R showed superior internal reliability. Additionally, the OCI-R measures six (to the MOCI's four) obsessive–compulsive constructs, and uses a more sensitive response format (likert scale vs. categorical). It is recommended that the OCI-R be employed as the self-report instrument of choice for assessing obsessive–compulsive pathology in those with AN. Copyright © 2010 John Wiley & Sons, Ltd and Eating Disorders Association.