OBJECTIVES: To explore the impact of severity and rate of decline of cognitive impairment on mortality within Alzheimer’s disease (AD) patients. METHODS: Data on AD patients were from CERAD (Consortium to Establish a Registry for Alzheimer’s Disease), a US multicenter, longitudinal study.  Patient demographics, cardiovascular and other co-morbidities, activities of daily living (ADL), cognitive impairment measured by the Mini-Mental State Examination (MMSE) and Clinical Dementia Rating (CDR), and survival status were obtained at study entry (baseline), and at annual follow-up visits.  Study survival time was modelled as a function of patient baseline demographics, cognitive impairment at entry and its decline over time using parametric survival models and semi-parametric Cox proportional hazard model (PH), with decline of cognitive impairment as a time-dependent covariate. RESULTS: 1010 patients with non-missing survival data were included in the analysis. At baseline, men, age and cognitive impairment significantly increased the mortality risk (p<0.001 for all). Additionally, patients progressing to a more severe stage of cognition over time had significantly higher mortality risk (p<0.001), with a hazard ratio for CDR change over time versus baseline as a time dependent covariate of 1.35 [95% CI: 1.21-1.51].  CDR and MMSE scores were highly correlated; results based on models using either CDR or MMSE were mostly consistent. Results were confirmed by sensitivity analyses with varying modelling assumptions. CONCLUSIONS: Severity and rate of decline of cognitive impairment are both strong predictors of survival in AD, additional to the usual demographic and clinical patient characteristics.  These findings suggest that treatments delaying the progression of AD patients to more severe stages of cognitive impairment might positively impact overall survival.