A recent study used data from two Phase III clinical trials of rivastigmine efficacy and safety to model rivastigmine s effect on the progression of Alzheimer s disease (AD). In these trials, a patient s AD status is gauged by measuring cognitive function using the mini-mental state exam (MMSE). The hazard model developed in this study has been used to estimate disease stage-specific savings in direct cost of caring for AD patients resulting from treatment with rivastigmine. OBJECTIVES: We refined this model to estimate precise MMSE score-specific savings and investigate the distribution of cost savings across direct and indirect costs of caring for AD patients. METHODS: MMSE score-specific estimates of AD progression in both untreated and treated patients were combined with both MMSE score-specific estimates of direct and indirect costs of AD, and estimates of the probability of institutionalization from previous studies. We estimated potential savings due to rivastigmine treatment in direct, indirect, and total costs of caring for AD patients in the US. We analyzed the relative magnitudes of these cost savings across MMSE scores for three treatment time horizons. RESULTS: As a percentage of total gross cost savings, savings in indirect costs are greatest for both mild and moderate patients during the first 6 months of treatment. After 2 years of treatment, gross direct cost savings make up the majority of overall cost savings for both mild and moderate patients. CONCLUSIONS: Decreases in likelihood of institutionalization resulting from treatment appear to be driving these results. Results clearly demonstrate long-term savings of early initiation of rivastigmine treatment.