The authors examined agreement between patients' utilities and importance rankings and clinicians' judgments of these assessments using a multiattribute model representing 6 aspects of health states potentially associated with localized prostate cancer. Patients were interviewed individually shortly after diagnosis and at a follow-up visit to obtain time-tradeoff utilities for 4 health states, including current health, and importance ranks of the 6 attributes. Their clinicians independently provided views of what utilities and importance ranks would be in the patient's best interest. Using patient-clinician pairs as the unit of analysis, the authors discovered that only about 50% of the correlations across 4 health states were high enough (.80) to be acceptable for clinical use for substituted judgment. Their conclusion: Clinicians should recognize that their judgments of the utility of health states associated with localized prostate cancer may not correspond closely with those of the patient. Copyright 2005 by the American Psychological Association.