I argue in this paper that the use of a global clinical impression of change as a primary measure in dementia drug trials is appropriate, given that we do not yet understand clearly enough what Alzheimer's disease looks like as it gets better.