This study aimed to validate the two total scores (TS-I and TS-II) of the Consortium to Establish a Registry for Alzheimer Disease neuropsychological battery (CERAD-NP) for a large elderly population including mild cognitive impairment (MCI) and dementia patients with various etiologic backgrounds. The authors also investigated whether the addition of frontal-executive function score can improve the discrimination accuracy of the total scores for dementia and MCI. One thousand three hundred thirty-six normal comparison (NC), 583 dementia (420 AD, 111 non-AD dementia, and 52 mixed AD with non-AD dementia), and 250 MCI (223 amnestic and 27 nonamnestic MCI) individuals living in the community were included (all aged 60 years and older). Both TS-I and TS-II were highly correlated with other global cognitive and functional scales. Both total scores showed, though modest, superior NC versus MCI discrimination ability to Mini-Mental State Examination (MMSE). Their discrimination ability for NC versus dementia was excellent and significantly better, especially in discriminating very mild dementia, than MMSE. The addition of frontal-executive test score to TS-I or TS-II did not make a significant improvement in dementia or MCI discrimination ability. Both of them also showed higher test-retest and interrater reliability than MMSE or any individual neuropsychological tests in the CERAD-NP. These results strongly support the validity and usefulness of CERAD total scores for early detection and progression monitoring of MCI and dementia in clinical and research settings.