Abstract Background: Many people with dementia die in long-term care settings. These patients may benefit from a palliative care goal, focused on comfort. Admission may be a good time to revisit or develop care plans. Objective: To describe care goals in nursing home patients with dementia and factors associated with establishing a comfort care goal. Design: We used generalized estimating equation regression analyses for baseline analyses and multinomial logistic regression analyses for longitudinal analyses. Setting: Prospective data collection in 28 Dutch facilities, mostly nursing homes (2007-2010; Dutch End of Life in Dementia study, DEOLD). Results: Eight weeks after admission (baseline), 56.7% of 326 patients had a comfort care goal. At death, 89.5% had a comfort care goal. Adjusted for illness severity, patients with a baseline comfort care goal were more likely to have a religious affiliation, to be less competent to make decisions, and to have a short survival prediction. Their families were less likely to prefer life-prolongation and more likely to be satisfied with family-physician communication. Compared with patients with a comfort care goal established later during their stay, patients with a baseline comfort care goal also more frequently had a more highly educated family member. Conclusions: Initially, over half of the patients had a care goal focused on comfort, increasing to the large majority of the patients at death. Optimizing patient-family-physician communication upon admission may support the early establishing of a comfort care goal. Patient condition and family views play a role, and physicians should be aware that religious affiliation and education may also affect the (timing of) setting a comfort care goal.