The purpose of this study was to compare patient and proxy (physician and nurse) assessments of symptoms in advanced cancer patients. The sample consisted of 49 patients with advanced cancer admitted to an acute palliative care unit. Three independent assessments were completed for each patient on two occasions within 11 days of admission. On each occasion, symptoms were rated independently by the patient and two proxies (treating physician and nurse), using the Edmonton Symptom Assessment System (ESAS). The ESAS is a nine-item visual analogue scale (VAS) for assessing pain, activity, nausea, depression, anxiety, drowsiness, appetite, well-being and shortness of breath. Symptom ratings were compared using a repeated-measures ANOVA procedure and correlations. Average physician ratings were generally lower than average patient ratings for both occasions. Average nurse ratings agreed more closely with patient ratings, with a trend towards lower ratings on occasion 1 and higher ratings on occasion 2. There was a significant rater (person rating the effects) effect (P < 0.01) for three of the nine symptoms: physicians rated drowsiness, shortness of breath and pain significantly lower than patients. For drowsiness and shortness of breath, these differences were clinically relevant, representing a difference of more than 12 mm on a 100-mm VAS. The accuracy of assessments amongst those rating the symptoms did not improve over time. Proxy assessments of symptom intensity, particularly by physicians, were significantly lower than patient assessments for three of the nine symptoms. Further research regarding the reliability of patient and proxy assessments is needed to assess and manage symptoms in advanced cancer effectively.