The status of Mild Cognitive Impairment (MCI) as a valid construct is controversial. The term encompasses people with heterogeneous clinical profiles, and invites sub-classifications that still require validation. Still, much evidence suggests that, properly selected, many people with MCI--especially Amnestic MCI--are at a high risk of dementia. This paper considers the validity of the construct of MCI as a high-risk state for progression and a target for treatment. We conclude that the status of MCI as an entity remains controversial. On the one hand, it can be argued that the careful section of cases at high risk of developing dementia means that it is a valid target, with the goal being the prevention of dementia. Advocates of this view see a linear progression that they are trying to arrest, but studies have yet to show that this can be done. On the other hand, it can be argued that the patients who progressed did not develop dementia: they actually had a very early form of it. By this view, people without the progressive form will be needlessly exposed to antidementia drugs, and the others should be treated anyway. Why some people progress and others do not is not clear, but the variable rates of progression--between clinic-based and population-based samples and between very similar clinical trials with slightly different inclusion criteria--suggests that MCI is a heterogeneous entity. The phenomenon of slowing or non-progression itself should be investigated, and such investigations likely should extend to people now classified as having mild dementia.