As the body of knowledge about Alzheimer disease (AD) expands, more evidence is appearing to suggest the existence of different subtypes. The delineation of subtypes is important for a variety of reasons, and there are probably several ways to identify subtypes. One crude but functionally promising distinction is between patients with early onset of AD and those with late onset. In theory, such a distinction appears to be a simple one, but critical methodologic issues are involved in this dichotomy, because a key aspect is estimating the rate of progression of the disease in different subtypes. This article discusses potential problems with various measures of deterioration and illustrates how the application of a novel technique-the trilinear model-can address many of these problems and provide additional valuable information. The trilinear model is especially well suited for use in pharmacologic studies because of the additional insight that it can provide.