Background: Caregivers of older adults with advanced cancer often have a different understanding of the patient’s prognosis compared with their oncologist. Among patients, accurate prognostic awareness is associated with greater depressive symptoms, except when patients utilize more adaptive coping skills. We examined the relationship between caregiver-oncologist prognostic concordance and caregiver depressive symptoms and explored whether this relationship differed by caregiver mastery, the capacity to cope, adjust, and adapt to problems. Methods: We utilized data from a national geriatric assessment cluster-randomized trial (URCC 13070: PI Mohile) that recruited patients aged ≥70 with incurable cancer considering any line of cancer treatment at community oncology practices, their caregivers, and oncologists. At enrollment, caregivers and oncologists estimated the patient’s prognosis (0-6 months, 7-12 months, 1-2 years, 2-5 years, > 5 years); same response was considered concordant. Caregivers completed Ryff’s mastery subscale (range 7-35, higher is better) at enrollment and depression screen (the Patient Health Questionnaire (PHQ)-2 (range 0-6) 4-6 weeks later. To assess the association of prognostic concordance with caregiver depressive symptoms, we used generalized estimating equations in models adjusted for cancer type, study arm, practice sites, and caregiver demographics. We then assessed moderation effect of caregiver mastery on this association. Results: Among 410 caregiver-oncologist dyads, mean caregiver age was 66.5, 75% were female, and 26% were caregivers of patients with lung cancer. Mean mastery score at enrollment was 27.6 (SD 4.7) and 19% screened positive on PHQ-2 at week 4-6. Among dyads who provided response (N = 370), 28% were concordant. Prognostic concordance was associated with higher caregiver depressive symptoms (β = 0.30; p = 0.04). Significant moderation effect was found between concordance and mastery for caregiver depressive symptoms (p = 0.02). Among caregivers with low mastery ( < median), prognostic concordance was associated with higher depressive symptoms (β = 0.68; p = 0.003). Among caregivers with high mastery (≥median), concordance was not associated with depressive symptoms (β = -0.06; p = 0.67). Conclusions: There is a need to study how prognostic understanding might lead to depression in at-risk caregivers. Interventions targeting caregiver prognostic understanding need to consider its relationship with depressive symptoms, while seeking to increase caregiver mastery.