Alzheimer's disease (AD) is associated with an increased mortality in comparison with aged control populations. The relationship between the clinical and the temporal course of AD has not been well studied over significant intervals. Community-residing patients with probable AD (N = 103, 42 men, mean age = 70.2 +/- 8.0 years) were studied at baseline on demographic and clinical variables, including measures of global deterioration (Global Deterioration Scale; GDS), mental status and cognition (e.g., Mini-Mental State Examination; MMSE), and functional impairment (Functional Assessment Staging; FAST). Baseline characteristics included a GDS range of Stage 4, 5, or 6 (38.8%, 39.8%, and 21.4%, respectively) and a mean MMSE score of 15.4 +/- 5.6. The mean follow-up interval was 4.6 +/- 1.4 years. Follow-ups were done blind to baseline measures and when necessary were conducted in residential and nursing home settings. Of locatable subjects (n = 95, 92%), 30 (31.6%) were decreased. Survivors (n = 65) had a mean GDS stage of 6.2 +/- 0.9 and a mean MMSE score of 5.1 +/- 6.9; 51% had MMSE scores of 0. Increased age and male gender, but not baseline clinical dementia variables, increased the risk of death (ps < .01). Change in clinical variables correlated significantly with time elapsed (r = .32, p < .05, for MMSE change, to r = .48, p < .001, for GDS change). Significant variance in temporal change (i.e., time elapsed) was accounted for by change in two of the five clinical measures studied (i.e., GDS and FAST; multiple r = .53). The results support previous estimates of mean duration of the GDS and FAST stages. For subjects with probable AD followed over approximately 5 years, clinical variables changed significantly over time in survivors. However, the majority of temporal variance in the course of AD remains unexplained.