Background: Quality of life (QOL) support is critical to high quality advanced cancer care and is associated with improved patient outcomes, but most patients report unmet supportive care needs. To understand supportive care provision in routine oncology, we examined the agreement, or congruence, between patients’ self-reported symptoms and topics discussed with oncologists over an extended interval of care. Prior research suggests that patient-provider rapport and overall QOL influence discussion of patient concerns, therefore we estimated effects of two predictors of interest: therapeutic alliance between patient and oncologist, and global QOL. Methods: We conducted a secondary analysis of data collected in the VOICE study. Patients with advanced cancer (n=196) completed measures of symptoms (e.g., pain, nausea, emotional concerns) and QOL, as well as an inventory of topics discussed with oncologists over the prior 3 months. Data were collected at quarterly intervals for up to 48 months, resulting in 901 paired reports of patients’ symptoms and discussions with oncologists. We constructed congruence as a multinomial outcome with four discrete categories: congruence (symptom reported/discussed), unaddressed concern (symptom reported/not discussed), surveillance (symptom not reported/discussed), and no supportive care concern (symptom not reported/not discussed). We then estimated effects of global QOL and therapeutic alliance using multi-level logistic regression. Results:Congruence was most common for pain (49.9%), while emotional concerns were most likely to be unaddressed (50.5%). Higher therapeutic alliance with oncologists was associated with lower odds of unaddressed pain (OR 0.94, CI 0.89-1.00) and emotional concerns (OR 0.95, CI 0.91-0.98). Higher global QOL was associated with higher odds of unaddressed emotional concerns (OR 1.11, CI 1.02-1.20). Congruence outcomes were stable over time. Conclusions: Oncologists may improve congruence by strengthening therapeutic alliance, by more consistently discussing emotional concerns, and by prioritizing topics for discussion according to patients’ needs.