OBJECTIVE: Three issues relevant to revising the DSM-III-R criteria for obsessive-compulsive disorder were examined in a field trial: 1) the requirement that symptoms of obsessive-compulsive disorder be viewed by the patient as excessive or unreasonable, 2) the presence of mental compulsions in addition to behavioral compulsions, and 3) ICD-10 subcategories. METHOD: The authors studied symptom patterns of obsessive-compulsive disorder as well as strength of obsessive belief among 431 patients with obsessive-compulsive disorder at seven hospital outpatient clinics. Two methods of subject selection were used: consecutive entry of everyone who contacted the clinics for evaluation of obsessive-compulsive disorder and entry of patients with obsessive- compulsive disorder who had continuing contact with the clinics since before the field trial and who were still symptomatic. Primary measures were the Yale-Brown Obsessive Compulsive Scale and face-valid questions about fixity of obsessive-compulsive beliefs. RESULTS: The large majority of patients were uncertain about whether their obsessive- compulsive symptoms were unreasonable or excessive, and most had both mental and behavioral compulsions. Results on the ICD-10 subcategories were equivocal. CONCLUSIONS: The present results converge with previous findings to indicate a broad range of insight among patients with obsessive-compulsive disorder. The DSM-III-R requirement for insight should be de-emphasized in DSM-IV, and mental rituals should be included in the definition of compulsions.