Hospitalization is a difficult experience, especially for patients with dementia. Understanding whether better continuity of care (COC) reduces hospitalizations can indicate interventions that might help curb hospitalizations. To estimate the causal impact of COC on hospitalizations and different reasons for hospitalization among community-dwelling older veterans with dementia. Population-based observational study using nationwide Veterans Health Administration data linked to Medicare claims in Fiscal Years (FYs) 2014-2015. To account for unobserved confounders we used an instrumental variable for COC-whether veteran changed residence by more than 10 miles. Community-dwelling veterans with dementia aged 66 and older, enrolled in Traditional Medicare (n=105,528). Bice-Boxerman Continuity of Care (BBC) index (0-worst to 1-best COC); binary indicators of any hospitalization for all causes, for ambulatory care sensitive conditions (ACSCs) and for reasons grouped by major diagnostic category. The mean BBC in FY 2014 was 0.32 (SD, 0.23). In FY 2015 43.3% of the cohort veterans were hospitalized. A 0.1 higher BBC resulted in 2.4% (95% confidence interval, 0.5%-4.4%) lower probability of hospitalization for all causes. BBC was not associated with hospitalization for ACSCs. Grouped by major diagnostic category, a 0.1 higher BBC resulted in 3.8% (95% confidence interval, 2.1%-5.4%) lower probability of hospitalization for neuropsychiatric diseases/disorders, with no impact on hospitalizations for circulatory, respiratory, infectious, kidney and urinary, digestive, musculoskeletal, and endocrine-metabolic diseases/disorders. Among community-dwelling older veterans with dementia, better COC resulted in less hospitalizations, and this effect was primarily due to less hospitalization for neuropsychiatric diseases/disorders but not hospitalization for ACSCs, or other hospitalization reasons.