This paper introduces the Composite International Diagnostic – Screener (CID–S) for mental disorders, a 12-item self-report questionnaire that is based on core diagnostic questions from the Composite International Diagnostic Interview for the assessment of DSM-IV and ICD-10 disorders (CIDI) for use in two-stage epidemiological designs. Data are reported relating to the performance of the CID–S in a two-stage study of a random sample of 1095 respondents, using a wide range of DSM-IV/CIDI diagnoses (somatoform, anxiety, depressive, other affective, and substance-use disorders). In the first stage, all subjects filled out the screening instrument. Seven to 30 days later trained clinical interviewers administered the computer-assisted full CIDI. The CID–S had an overall sensitivity of 85.3% (range for specific diagnoses 74.7% for alcohol use disorders to 100% for panic disorders) and negative predictive value (NPV) of 92.0% (range 91.0%–100%) for any current (12-month) diagnosis and similar values for lifetime disorders (80.7%, NPV: 85.1%). Likelihood for a DSM-IV diagnosis increased steadily from 10.5% among screen negatives, to 24.9% among those endorsing one screening item to 100% among those endorsing more than eight of the screening items. The performance of the individual CID–S items overall was good, except for three disorders: agoraphobia, somatoform disorders and substance use disorders. Overall, the CID–S was found to be a time-efficient diagnostic screening tool for most mental disorders (except for some somatoform and substance use disorders), when used in a two-stage design with a two to 12 week interval between investigations. Copyright © 1999 Whurr Publishers, Ltd.