Objectives: The Addenbrooke’s Cognitive Examination (ACE) is a common cognitive screening test for dementia. Here, we examined the relationship between the most recent version (ACE-III) and its predecessor (ACE-R), determined ACE-III cutoff scores for the detection of dementia, and explored its relationship with functional ability. Methods: Study 1 included 199 dementia patients and 52 healthy controls who completed the ACE-III and ACE-R. ACE-III total and domain scores were regressed on their corresponding ACE-R values to obtain conversion formulae. Study 2 included 331 mixed dementia patients and 87 controls to establish the optimal ACE-III cutoff scores for the detection of dementia using receiver operator curve analysis. Study 3 included 194 dementia patients and their carers to investigate the relationship between ACE-III total score and functional ability. Results: Study 1: ACE-III and ACE-R scores differed by ≤1 point overall, the magnitude varying according to dementia type. Study 2: a new lower bound cutoff ACE-III score of 84/100 to detect dementia was identified (compared with 82 for the ACE-R). The upper bound cutoff score of 88/100 was retained. Study 3: ACE-III scores were significantly related to functional ability on the Clinical Dementia Rating Scale across all dementia syndromes, except for semantic dementia. Conclusions: This study represents one of the largest and most clinically diverse investigations of the ACE-III. Our results demonstrate that the ACE-III is an acceptable alternative to the ACE-R. In addition, ACE-III performance has broader clinical implications in that it relates to carer reports of functional impairment in most common dementias. (JINS, 2018, 24, 854–863)