To determine if the occurrence of health outcomes following clinical laboratory testing can be used to identify types of laboratories that may be having higher than expected error rates. —Retrospective analysis of Medicare Part B outpatient claims, Part A hospitalization bills, and death records using a case-control study. —Medicare records from six carrier territories were sampled during the period 1985 through 1987. —A total of 14 755 Medicare patients receiving a prothrombin time test in either a physician office laboratory or a commercial laboratory. —Occurrence of a hospitalization for stroke or acute myocardial infarction, death, or no adverse outcome within 6 days of a prothrombin time. —In physician office laboratories where prothrombin time test volume is below 40 per month, the odds that a tested patient will experience a stroke or an acute myocardial infarction are up to 1.96 and 3.43 times greater, respectively, than for a similar patient tested in a commercial laboratory. Switching from one laboratory to another between successive prothrombin time tests increased the odds of a stroke or an acute myocardial infarction by 1.57 and 1.32, respectively. Patients in two states with strong laboratory regulatory programs had fewer adverse outcomes. —Examining patient outcomes subsequent to clinical laboratory testing may be a useful tool for clinical laboratory quality assurance. (JAMA. 1993;269:1030-1033). © 1993, American Medical Association. All rights reserved.