Use of a coagulation panel [prothrombin time (PT), partial thromboplastin time (PTT), thrombin time (TT) and fibrinogen], intended for evaluation of bleeding, tripled over 6 years, out of proportion to admissions, surgery, or transfusions. To determine whether the panels were ordered appropriately, we classified 28,737 sets of panel results into groups followed by chart reviews to determine typical patient histories. In 39% of panels, PT/PTT was normal. Prolonged PT occurred in 33% of results, due to liver failure (8%), warfarin (23%), and presumed vitamin K deficiency (69%). Prolonged PTT occurred in 34% of results and was primarily associated with long PT or lupus inhibitors. Prolonged PTT and TT (15% of panels) indicated heparin therapy. Fibrinogen was normal in 98% and low in 1.4%. Critical fibrinogen (below 100 mg/dl, 0.6% of panels) was associated with bleeding in 90% of patients. Only 8% of panel orders were clinically indicated based on patient history. Clinician interviews indicated many were unaware the panel included fibrinogen and TT. Interventions included an education program and an order form change. The education program had no effect on overall order volume or test selection. A later order form change made TT and fibrinogen a separate order. This reduced TT and fibrinogen testing by 90% without complaints or changes in blood transfusion statistics. We conclude that many coagulation test panel orders were not clinically indicated, that PT more often diagnosed vitamin K deficiency than bleeding risk, and that order-based restriction of testing was more effective than educational programs at introducing change in clinical test utilization.