Crafting care that is concordant with the patient's wishes in the context of serious illness requires clear, frank and sensitive communication focused on the issues of greatest importance to the patient. We aim to test a potentially powerful intervention - targeting physicians, patients and caregivers - to improve discussions regarding prognosis and treatment choices in advanced cancer. These discussions are critical to involving patients in their care, whether or not they wish to assume responsibility for making major health care decisions. Patients who have had these discussions are less likely to be misinformed about the illness, less likely to get unwarranted aggressive care and more likely to experience improved well-being. Yet, these crucial discussions frequently do not occur. As predicted by ecological theory, individually-focused interventions - targeting clinicians or patients, or using third parties - are of limited benefit in improving communication. Our approach, informed by ecological theory, is designed to optimize communication - and patient well-being - by aligning patient, caregiver and physician communication towards common goals. In this resubmission, we respond to PA-09-122: Research on Clinical Decision Making in People with or at Risk for Life-Threatening Illness and have new Specific Aims: To determine whether a combined intervention for patients/caregivers and physicians a) improves communication regarding prognosis and treatment choices in advanced cancer (primary outcome), b) improves patient well-being and c) affects health services utilization. Design is a two-site cluster RCT of a 2- component intervention designed to improve communication about prognosis and treatment choices in advanced cancer and promote patient participation in discussions regarding their care. Oncologists (N=30) will be randomized to usual-care control or the intervention: two in-office sessions providing individual tailored feedback on their communication skills. Patients (n = 300) with advanced cancer and their caregivers will be assigned to usual-care control or the intervention: pre-visit coaching and question-prompt lists to address the same communication goals as physicians. We have piloted both components of the intervention. Main outcome measures derived from audio-recorded patient-caregiver-oncologist visits will assess a) interactive communication, including responding to patients' concerns; informing patients about treatment choices; balanced framing of prognoses; and engaging patients to participate in decisions. We will also assess b) communication 1 week following the visit and patient well-being (psychological distress, quality of life, sense of peace, and quality of death) at study entry and every 3 months using patient and caregiver surveys; and c) utilization of aggressive interventions in the last week of life and palliative care and hospice consultations following study enrollment using chart audit. Significance: The proposed study addresses critically important barriers to effective communication in the context of life-limiting illness. It is the first to test a novel communication intervention targeting physicians, patients and caregivers, and it is scalable for dissemination.