It is increasingly difficult to choose the proper tests to investigate an abnormal laboratory result, delaying the time to reach the correct diagnosis and increasing the cost of care. A wrong choice may also lead to clinically significant errors, which can be life threatening. To show how errors in test selection occur in routine medical practice, the authors describe an algorithm to evaluate patients with a prolonged activated partial thromboplastin time (PTT) and a normal prothrombin time. This exercise challenges the commonly held belief that errors in laboratory utilization occur primarily with esoteric and/or genetic testing, rather than in patients with common abnormalities such as a prolonged PTT. © 2012 Wiley Periodicals, Inc.