The present report analyzes the agreement between the interview and the self-report formats of the instruments Structured Clinical Interview for Social Anxiety Spectrum (SCI-SHY) and Structured Clinical Interview for Obsessive Compulsive Spectrum (SCI-OBS), already validated, in three psychiatric patient samples and controls. Thirty patients (10 with obsessive-compulsive disorder [OCD], 10 with social anxiety disorder [SAD], 10 with recurrent unipolar depression in remission) and 20 control subjects (10 university students, 10 ophthalmologic patients) were assessed using the SCI-SHY, the SCI-OBS, and the self report version of the two instruments. Agreement between the two versions was very good for the seven SCI-OBS domains (with intraclass correlation coefficients [ICCs] ranging from 0.80 to 0.96) and the four SCI-SHY domains (ICCs from 0.74 to 0.90). When items were analyzed individually, subjects tended to under-report some phobia-related problems in the interview. The total number of items endorsed in the SCI-SHY, but not in the SCI-OBS, was affected by the order of administration: when the SCI-SHY interview was administered first, subjects reported a median of five more symptoms; when the self-report was administered first, there was no significant difference in the number of symptoms endorsed in the two formats. However, this difference is not clinically important, given the large number of items comprising the instruments, and might be explained by the fact that subjects are likely to overemphasize occasional symptoms or behaviors when they are asked by the interviewer to answer a long series of [ldquo ]new[rdquo ] questions as accurately as possible. Given the high agreement between domain scores in the two formats of the instruments and the fact that scores are virtually identical when the self-report is administered first, we recommend the use of the self-report versions in clinical and research settings.