Psychiatric disorders in people with learning disabilities are often undetected because the possibility of a mental health problem is not considered. Instruments used for detection of psychiatric disorders in the general population have not been shown to be appropriate for this population. At present, the instruments with the best information about satisfactory validity in this population are the PASADD interview as a diagnostic instrument for adults, the Reiss Screen as a detection instrument in . adults, and the Developmental Behaviour Checklist in children and adolescents. Aims: to ascertain the validity of screening questionnaires in detecting psychiatric disorders among adults with learning disabilities. To compare the characteristics of a new screening questionnaire (the PASADD Checklist) with an existing instrument (The Reiss Screen for Maladaptive Behavior). To make an estimate of the prevalence of psychiatric disorder in a population of adults with moderate and severe learning disabilities. To compare the prevalence of psychiatric disorder in this population with that of the general population. Design: Cross-sectional descriptive study with a 25% random sample drawn from a register of adults using health and social services for people with learning disabilities in a metropolitan borough in the North-West of England in 1996-7, using a semi-standardized interview, the Psychiatric Assessment Schedule for Adults with Developmental Disabilities (PAS-ADD) as a ’gold standard’ to undertake Receiver Operating Characteristic analysis, and to calculate the best cutoff score of a new instrument, the PASADD Checklist, and the Reiss Screen. A further 12.7% random sample was added to measure prevalence. Results: The response rate was 84%. 146 adult users of learning disability services took part, with their main carers who acted as informants. 18 individuals (12.3%) were diagnosed using the ICD-lO Diagnostic Criteria for Research. DSM-IV Axis I diagnoses were made for 38 individuals (26.0%). Internal consistency for the PASADD Checklist was good. Four scoring methods were examined, and the one with the best screening performance was the simplest -counting any positive response to an item as 1 and adding up the number of symptoms reported. The best agreement with the PASADD interview was obtained by using information only from the main carer. Correlation between the Checklist score and the severity of psychiatric disorder was fair. The best cutoff score was 3 (Detection Rate = 100%, False Positive Rate = 54% when compared with the ICD-10 Diagnostic Criteria for Research; DR = 91 %, FPR = 47% compared with DSM› IV diagnosis). Conclusion: Psychiatric disorders with recent symptoms sufficient to make a diagnosis were found in 26% of adults with learning disability in a survey of service users. The PASADD Checklist is acceptable to carers, and gives satisfactory screening performance with a simple scoring method.