OBJECTIVES: According to 2013 American Geriatrics Society Updated Beers Criteria, paroxetine has strong anticholinergic properties than other Selective Serotonin Reuptake Inhibitors (SSRIs). Such anticholinergic effects may lead to adverse cognitive outcomes.  This study examined the risk of dementia associated with the use of paroxetine versus other SSRIs.    METHODS: A retrospective cohort study was conducted using 2007-2010 Medicare claims data, and included nursing home residents > 65 years with depression. The study focused on incident SSRI users who did not have dementia in 2007 (baseline). Patients were included if they had continuous coverage for Medicare Parts A, B and D and no HMO coverage during the one year baseline and 2 years of follow up or until death. The primary outcome of this study was time to dementia diagnosis. SSRIs were classified as paroxetine and others. Cox proportional hazards regression was conducted to evaluate the risk of dementia with the use of paroxetine versus other SSRIs. RESULTS: The study cohort consisted of 19,050 elderly nursing home residents with depression. Among SSRI users, 1,716 (9.01%) received paroxetine and 17,334 (90.99%) received others. Since proportional-hazard assumption was violated, the extended Cox hazard model involving Heaviside function was used to evaluate the dementia risk. The extended model revealed that paroxetine users had 66% [Hazards Ratio, HR, 1.66; 95% Confidence Interval (CI), 1.03-2.67] higher risk for dementia than other SSRIs users after 390 days of treatment. However, the dementia risk did not vary within 390 days of SSRI use. Other factors positively associated with dementia risk were age, male gender, and non-White race. CONCLUSIONS: Paroxetine use was associated with a time-varying increase in risk of dementia among depressed elderly nursing home residents. There is a need to optimize anticholinergic medication use in this population as depression is an independent risk factor for dementia.