The clinician must always optimize the patient's chance for the best possible outcome, but the complex task of deciding which diagnostic study to perform and which therapeutic agent to administer is usually carried out using imperfect information. These critical decisions are usually made in a covert fashion without explicit consideration of the multiple factors that impact on the outcome. Decision analysis is a quantitative method for making such diagnostic and therapeutic judgments, incorporating both probabilistic data and value judgments in the analysis of clinical problems. This article presents a brief overview of decision analysis and a detailed illustration of its use in an actual complex clinical situation. The clinical problem described consists of the management of suspected pulmonary embolism in a patient with a high risk of complications from both pulmonary arteriography and long-term anticoagulation. The example chosen also points up the virtue of decision analysis in considering prospectively both the information content and the attendant risks of contemplated diagnostic procedures.