Background Caregiver perceived autonomy support by the oncologist is important for caregiver well‐being and may be affected by the patient's survival. We determined the association of caregiver‐oncologist discordance in patient's life expectancy estimates with perceived autonomy support over time and whether the association differed by patient survival status. Materials and Methods We used data from a geriatric assessment cluster‐randomized trial (URCC 13070) that recruited patients aged at least 70 years with incurable cancer considering or receiving treatment, their caregivers, and their oncologists. At baseline, caregivers and oncologists were asked to estimate patient's life expectancy (0–6 months, 7–12 months, 1–2 years, 2–5 years, and >5 years; any difference in response was considered discordant). At 4–6 weeks, 3 months, and 6 months, caregivers completed the Health Care Climate Questionnaire (HCCQ), which measured perceived autonomy support by the oncologist. Generalized estimating equation modeling was conducted to assess the association of baseline caregiver‐oncologist discordance with longitudinal HCCQ scores, stratified by patient 6‐month survival status. Results Discordant life expectancy estimates were present in 72.0% of dyads. In multivariate analyses, caregiver‐oncologist discordance in patient's life expectancy estimates was associated with higher caregiver HCCQ scores. In stratified analysis, caregiver‐oncologist discordance was associated with lower caregiver HCCQ scores (β = −3.46; 95% CI, −4.64 to −2.29) among patients who died within 6 months but with higher caregiver HCCQ scores (β = 1.33; 95% CI, 0.63–2.04) among patients who survived beyond 6 months. Conclusion Interventions aimed at mitigating discordance need to consider its association with caregiver perceived autonomy support and patient's survival in order to better inform caregiver expectations. Implications for Practice Among patients who died within the first 6 months, caregivers who estimated a different length of life for the patient compared with oncologists were more likely to report lower support from the oncologist, whereas the opposite relationship was seen within patients who survived beyond the first 6 months. When designing interventions to improve caregiver understanding of the patient's prognosis, its relationship with caregiver‐perceived support and patient's survival needs to be considered. Prognostic discordance refers to a discrepancy in estimates of prognosis by two individuals. This article evaluates the association between caregiver‐oncologist prognostic discordance regarding a patient's life expectancy estimates and caregiver perceived autonomy support.