Phenomenological differences between delirium superimposed on dementia (DsD) versus delirium in the absence of dementia (DaD) remain poorly understood. We aimed to identify phenomenological differences in delirium symptoms (cognitive and non-cognitive) and compare delirium recovery trajectories between DsD and DaD. We conducted a prospective observational study on individuals admitted to the Geriatric Monitoring Unit (GMU), a five-bed unit specializing in managing older adults with delirium, between December 2010 and August 2012 (n = 234; mean age 84.1 ± 7.4). We collected data on demographics, comorbidities, severity of illness, cognitive and functional scores, and number of precipitants. Cognitive status was assessed using locally validated Chinese Mini-Mental State Examination (CMMSE) and delirium severity assessed using Delirium Rating Scale-Revised-98 (DRS-R98). Delirium disease trajectory was plotted over five days. DsD patients had a longer duration of delirium with slower recovery in terms of cognition and delirium severity scores compared with DaD patients (0.33 (0.0-1.00) vs. 1.0 (0.36-2.00) increase in CMMSE per day, p < 0.001, and 1.49 ± 1.62 vs. 2.63 ± 2.28 decrease in DRS-R98 severity per day, p < 0.001). When cognitive and non-cognitive sub-scores of DRS-R98 were examined separately, we observed steeper recovery in both sub-scores in DaD patients. These findings remained significant after adjusting for significant baseline differences. Our findings of slower cognitive symptom recovery in DsD patients suggest cognitive reserve play a role in delirium syndrome development and recovery. This merits further studies to potentially aid in appropriate discharge planning and to identify potential pharmacological and non-pharmacological cognitive interventions for hospitalized older persons with delirium.