BACKGROUND: An observer-rated screening questionnaire for dementia for people with intellectual disabilities (ID), DSQIID, was developed in the UK. So far, the German version has not yet been validated in adults with ID. AIMS/METHODS: We validated a German version of DSQIID (DSQIID-G) among adults with ID attending a German clinic. PROCEDURES/OUTCOMES: DSQIID-G was completed by the caregivers of 104 adults with ID at baseline (T1), 94 at six months (T2) and 83 at 12 months (T3). A Receiver Operating Curve (ROC) was used to determine the total DSQIID-G cutoff score for the best fit between sensitivity and specificity. RESULTS: Sixteen of the 104 participants at T1 (15%) received a diagnosis of dementia. At T1, the scores among the non-dementia group ranged from 0 to 33 (mean: 6.7; SD: 7.65), and the dementia group ranged from 3 to 43 (mean: 22.12; SD: 11.6). The intergroup difference was statistically significant (W: 158; p < .001) (AUC:.89). A total score of 9 provided the best fit between sensitivity (.94) and specificity (.72). CONCLUSIONS AND IMPLICATIONS: DSQIID-G total score can discriminate between dementia and non-dementia cases in adults with ID. A lower cutoff score with a higher sensitivity is desirable for a screening instrument.