OBJECTIVES:  To evaluate the relative likelihood of adverse events, including relapse of psychosis, stroke and death, associated with continued “off-label” use of risperidone in elderly patients with dementia compared to those who discontinued use of the drug after four months. METHODS:  This population-based cohort study analyzed long-term care (LTC) patients in a large nationally representative and statistically de-identified administrative claims database. The sample consisted of dementia patients aged 65+ treated with risperidone between 2008 and 2011. Patients were identified by the existence of a new prescription fill for risperidone (treatment naïve) and a diagnosis for dementia (without bipolar or schizophrenia) within one year prior to the index fill. Patients were followed from 30 days up to 24 months after four consecutive months of risperidone use to assess the relative risk of death, stroke and time to relapse of psychosis. RESULTS:  The study population included 4,672 patients with dementia that were prescribed risperidone between 2008 and 2011 (female = 65.3%, mean age = 82.3 [± 5.6]). Findings reveal significantly higher incidence of psychosis in patients with continued use compared to those who discontinued use of risperidone (22.0% vs. 12.0%, p<0.01), which contradicts clinical trials. CONCLUSIONS:  Contrary to existing literature, this study found that elderly LTC dementia residents who discontinued use of risperidone after four months had a lower risk of psychosis as compared to those who continued use of the drug. This large claims-based analysis demonstrates the value of validating clinical trial outcomes in real-world settings. Additional research is needed to further investigate the safety of long-term “off-label” use of risperidone in elderly LTC patients with dementia.