Objectives: Caffeine is known to improve concentration and reduce fatigue in healthy adults, but high doses may induce anxiety and agitation. Because the effects of caffeine in elderly people with dementia are unknown, this study explores the relation between caffeine and behavioral symptoms in a group of elderly patients with dementia. Design: An observational pilot study. Setting: A dementia special care unit of a Dutch nursing home. Participants: A total of 29 elderly patients with dementia. Measurements: Behavioral symptoms were measured with the NPI-NH, and sleep and caffeine consumption were measured using questionnaires. Results: A significant relation was found between the total amount of caffeine consumed during the day and apathy [Kendall's tau (KT) -0.287 p=0.03], and the number of times that participants got up at night (KT 0.462; p <0.01). The amount of caffeine consumed after 6 p.m. was also significantly related to the number of times participants got up at night (KT 0.436; p <0.01). Multilevel analysis showed caffeine to be negatively correlated with aberrant motor behavior [b=-0.47 (0.22), Wald (461)=-2.12, p=0.03] and apathy [b=-0.88 (0.45), Wald (461)= -1.96, p=0.05], and showed a significant relation between caffeine consumption after 6 p.m. and the number of times participants got up at night [b=0.48 (0.22), Wald (461)= 2.20, p=0.03]. Conclusion: This study established an association between caffeine consumption and behavioral symptoms in elderly patients with moderately severe dementia. Therefore, adjusting caffeine consumption could be part of an interdisciplinary approach to behavioral symptoms, particularly when aberrant motor behavior, apathy or sleeping difficulties are involved. These results indicate that further research on the effects of caffeine on behavioral symptoms in dementia is warranted.