When screening for depression in glioma patients, the utility of proxy carer report is unknown. We studied how patients and proxies differed in the frequency, severity and agreement of reported depressive symptoms, the external validity of these reports, and whether patient-proxy agreement was associated with cognitive function. This was a cross-sectional study within a prospective cohort study of depression in glioma. Eligible patients were adults with a new diagnosis of cerebral glioma whose cohabiting partners chose to attend study interviews. Patients completed the Patient Health Questionnaire-9 (PHQ-9, maximum score 27) to screen for major depressive disorder. Proxies independently completed the PHQ-9 'for the patient'. A structured clinical interview for MDD was then given. From 55 couples attending, 41 participated (74 %). Patient-proxy total PHQ-9 score differed by 3 or more points in 26/41 cases (63.4 %). Disagreement within dyads ranged from -7 to +10 points. Proxies observed more individual depressive symptoms than patients reported (mean 2.7 vs 1.8 symptoms respectively, p = 0.013, Wilcoxon Rank Sum Test), and a greater severity of symptom burden (mean PHQ-9 score 8.4 vs 6.8 respectively, p = 0.016, Wilcoxon Rank Sum Test). Proxies were more reliable than patients on objective behavioural symptoms of depression. Dyadic agreement was not associated with severity of patient cognitive impairment. There was frequent disagreement between glioma patients and proxies reports of depressive symptoms. Proxies reported more depressive symptoms than patients, and were more reliable when reporting observable behavioural symptoms. When diagnosing depression in glioma, collateral history should be obtained.