The use of healthy individuals acting as simulated patients for the purpose of clinical teaching is discussed. The term "standardized patients" is used to refer to training the individual to present a standard, repeatable stimulus. Some evidence suggests that simulated patients possess high fidelity (i.e., closely approximate real patients). The reliability of scores based on simulated patients depends on the scoring method. Inter-rater reliability based on a single patient encounter ranges from acceptable to very high, depending on the degree of objectivity present in the scoring system. The question of validity, as opposed to fidelity, is a correlational one: i.e., do individuals who perform well or poorly on simulated patient encounters also perform well or poorly on other measures of competence? Evidence appears to indicate that scores derived from simulated patient encounters primarily reflect interpersonal skills, and that these scores are unrelated to content knowledge. A consistent finding has been that students taught with simulated patients out-performed those taught with real patients. A major barrier to the extensive use of standardized patients is feasibility (cost of training and application, use of faculty resources, and acceptability by faculty). (SW)