Till now, no specific diagnostic instruments to detect personality disorders in the elderly are available. The aim of our study was to contribute to the construction of a reliable en valid instrument for the detection of personality disorders in older adults. Therefore, a draft version of the instrument was constructed both for the patient and the informant. Both the patient and the informant version of the instrument included 52 items with the same content. These items concerned Habitual behaviour (HAB), Biographical information (BIO) and Observation of actual behaviour (OBS). 159 clients of 60 years and older of a Dutch ambulatory mental health care organisation were assessed with the screening instrument. In 96 of the 159 outpatients one or more informants were included. The results indicate that the internal consistency of seven HAB items, nine BIO items and five OBS items was moderate to good with regard to both the patient and informant instrument. Test-retest reliability of the 21 items of the patient instrument was moderate with regard to HAB, excellent with regard to BIO and moderate with regard to OBS. Interrater reliability of OBS was also moderate. The criterion validity (criterion: DSM-IV TR Axis II classification) of HAB, BIO and OBS based on the 21 items of the patient instrument was fair and with regard to the same 21 items of the informant instrument insufficient. In conclusion, sixteen out of 52 potential items of the patient's draft version of the screening instrument can be used to detect a possible personality disorder. These sixteen items from the HAB and BIO scale are named the Gerontological Personality disorder Scale (GPS). Medical doctors, psychologists and nurses working in ambulatory mental health organisation can use the GPS as a resource during the diagnostic process. Whether the GPS can be used beyond mental health organisations, for example by general practitioners is a subject for future studies.