Objectives To elucidate the hypothetically different interactions between delirium and post-discharge prognostic indicators in elderly hospital inpatients with versus without dementia. Design Retrospective cohort study of claims data by Taiwan National Health Insurance beneficiaries between 2002–2013. Setting Records of public hospital admissions in the Taiwan National Health Insurance Research database. Participants Propensity-score matched subgroups of patients with delirium superimposed on dementia (n = 922) versus dementia alone (n = 922), delirium alone (n = 680) versus neither delirium nor dementia (n = 680). Measurements Mortality, emergency department visits, readmissions, and psychotropic drug use, within 30, 180, and 365 days of discharge, were analyzed using multivariate proportional hazards or logistic regression analyses. Results Delirium superimposed on dementia was not associated with increased post-discharge mortality, or emergency department visits, but significantly increased the risk of readmissions at 365-day follow-up (adjusted HR, 95% CI: 1.26, 1.01–1.56). However, delirium without dementia was significantly associated with increased post-discharge mortality, emergency department visits and readmissions at 180 days and 365 days (respective adjusted HRs: mortality, 1.63 and 1.79; adjusted ORs: emergency department visits, 1.89 and 1.81; readmissions, 1.47 and 1.53). Delirium in patients both with dementia and without, was associated with six-fold higher likelihood of in-hospital psychotropic drug use, and doubled post-discharge psychotropic drug usage. Conclusions The obvious association between in-hospital delirium and worsened long-term prognosis, irrespective of dementia, raises awareness to warrants proactive and multimodal prevention and intervention strategies. Furthermore, the mechanisms about different influence of delirium for patients with/without dementia need to be further explored.