The rate of cognitive decline in AD has been noted to vary significantly among patients. The ability to predict the probable rate of decline early in the disease process would be of great practical importance. Attempts to analyze early cognitive deficits to find patterns associated with rapid decline have met with limited success. This paper utilized a large sample of patients with a diagnosis of probable AD evaluated longitudinally in ongoing research at the ADRC at Baylor College of Medicine and a statistical procedure of profile analysis to assess the initial data for a pattern associated with rapid decline. The findings indicated that despite initial equality of MMSE scores, patients showing rapid MMSE decline at one year displayed significantly more impaired performance on neuropsychological measures at diagnosis. Discussion includes discussion on the use of the MMSE for tracking general cognitive function and the difficulties of ascertaining stable profiles for prediction.