This overview focuses on the cognitive transition between normal aging and dementia. Numerous studies indicate that individuals who will go on to develop dementia show cognitive deficits many years before the time at which a clinical diagnosis could be rendered. The degree of preclinical impairment is remarkably similar for tasks assessing episodic memory, executive functioning, and perceptual speed, consistent with the view that multiple brain alterations occur prior to clinical disease onset. Although most research in this area has dealt with Alzheimer disease (AD), several recent reports indicate that the pattern of preclinical impairment is very similar in the second largest dementia disorder, vascular dementia (VaD). This is important because currently the possibility for interventions to postpone disease onset is greater in VaD than in AD. Despite pronounced preclinical cognitive deficits in dementia, the performance distributions between cases and controls are largely overlapping, hampering the ability to identify high-risk individuals. To alleviate this problem, future research should evaluate hybrid models for the prediction of dementia. In such models, multiple indicators of cognitive functioning should be included along with markers from other domains that have been linked to subsequent dementia (such as brain imaging, genetics, and lifestyle variables). To the extent that these categories of variables add unique variance, classification accuracy will increase and the overlap in performance scores between incident cases and controls will decrease, thereby enhancing clinical usefulness. This approach would also facilitate the examination of interactive effects among classes of preclinical markers.