Background: Nursing home residents with dementia show a rapid decline in their ability to perform activities of daily living. This decline is linked to a greater care dependency, which is associated with a reduced quality of life. Care dependency is influenced by multiple predictors, yet current research often focuses on the contribution of a single or a small number of predictors of care dependency. Objectives: To examine the contribution of multiple predictors in predicting care dependency. Design: The present study analyzed baseline data from a 6-month double-parallel randomized controlled trial which examined the effect of three physical activity interventions on multiple outcomes. Setting: This study was conducted in eleven nursing homes in Bergen op Zoom, the Netherlands. Participants: In total, 85 nursing home residents with moderate to severe dementia were included in the study, of which 75 were included for analysis. Methods: Predictors considered were cognitive, physical, neuropsychiatric, demographic, and disease related factors. The outcome measure care dependency was assessed with the Care Dependency Scale and the Erlangen Test of Activities of Daily Living. Linear multilevel regression analyses were used to identify the most important predictors of care dependency. Results: Apathy, physical endurance, number of comorbidities, and global cognition were significant predictors of care dependency. The model explained 66% of the variance in care dependency. Global cognition was a significant predictor of ability to perform activities of daily living and explained 60% percent of its variance. Conclusion: The present study shows that multiple predictors (i.e., apathy, cognitive and physical abilities, and disease-related factors) contribute to predicting care dependency. Future research could focus on the effectiveness of multifactorial interventions to maintain the highest possible level of independence in nursing home residents with dementia.