Objectives: To investigate whether deficits in empathic accuracy (i.e., ability to recognize emotion in others) in patients with neurodegenerative disease are associated with greater depression in their caregivers. Design: Two cross-sectional studies. Setting: Academic medical center and research university. Participants: Two independent samples (N = 172, N = 63) of patients with a variety of neurodegenerative diseases and their caregivers; comparison group of healthy couples. Measurement: Patients' empathic accuracy was assessed in the laboratory using a novel dynamic tracking task (rating another person’s changing emotions over time) and more traditional measures (recognizing the emotion expressed in photographs of facial expressions and by characters in films). Caregivers completed self-report inventories of depression. Results: Lower empathic accuracy in patients was associated with greater depression in caregivers in both studies. In study 1, this association was found when empathic accuracy was measured using the dynamic tracking measure but not when measured using the more traditional photograph and film measures. In study 2, we found preliminary support for our theoretical model wherein lower empathic accuracy in patients is associated with increased caregiver stress (loneliness, strain, and burden), which in turn is associated with greater caregiver depression. Conclusions: Caring for a patient with deficits in empathic accuracy is associated with greater loneliness, strain, and burden for caregivers, and increased depression. Caregivers may benefit from interventions designed to compensate for the stress and interpersonal loss associated with patients' declining empathic accuracy.