A major goal of palliative medicine is to control symptoms that interfere with quality of life. Identification of symptoms that occur together (cluster) may aid in symptom management, resulting in greater therapeutic benefit to the patient. An analysis of 25 symptoms from 922 patients with advanced cancer was undertaken to determine if symptom clusters could be identified. Cluster analysis was done using an agglomerative hierarchical method with average linkage; the absolute value of the correlation between pairs of symptoms was used as the measure of similarity. A correlation of >or=0.68 was used to define the final clusters. Seven clusters were identified: (1) fatigue: anorexia-cachexia; (2) neuropsychological; (3) upper gastrointestinal; (4) nausea and vomiting; (5) aerodigestive; (6) debility; (7) pain. Recognition of symptom clusters should help understand symptom pathophysiology and target therapies that perhaps can be used to relieve multiple symptoms in that cluster. This could result in improved quality of life for patients with advanced cancer and perhaps reduce polypharmacy, lessen drug side effects, and have pharmacoeconomic benefits.