Background: In Taiwan, the causes of death and related factors in the oldest old people with Alzheimer disease (AD) are not well characterized. We investigated the factors associated with mortality in the oldest old patients with newly diagnosed AD admitted to a long-stay residential facility. Methods: We performed a prospective study of newly diagnosed AD patients at a veterans' home between 2012 and 2016. At admission, all eligible participants received a comprehensive geriatric assessment, including demographic variables, lifestyle habits, cognitive evaluations, medical conditions (comorbidities, Age-Adjusted Charlson Comorbidity Index score, and polypharmacy), nutritional status evaluated by the Mini Nutritional Assessment-Short Form and body mass index (BMI), and global functional status. A Cox proportional hazards model was used to evaluate the predictive values of clinical parameters for all-cause mortality. Results: The cohort comprised 84 newly diagnosed AD patients (mean age 86.6 ± 3.9 years) with a mean follow-up period of 2.1 ± 1.2 years. The overall median survival was 3.5 years from the time of AD diagnosis (95% confidence interval, 3.1-3.9 years). BMI was significantly lower in the deceased group than in the alive group (20.7 ± 2.9 vs. 22.6 ± 3.4, p = 0.023). Logistic regression demonstrated that the clinical parameters significantly associated with mortality were high global comorbidity, low nutritional status (malnutrition and underweight), and impaired physical function at the time of AD diagnosis. Conclusion: Comorbidity burden, nutritional status, and physical functional status at the time of dementia diagnosis are important contributors to poor outcome in the oldest old. Efforts to control concurrent chronic disorders, nutritional interventions, and physical independency as a long-term care strategy for dementia may provide survival benefit.