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.