Each year in the United States, millions of units of fresh frozen plasma and platelets are transfused. A substantial proportion of these units are given to nonbleeding patients to "decrease the risk" of bleeding at the time of invasive procedures. The decision to transfuse is frequently triggered by an elevated prothrombin time (PT; international normalized ratio ) or activated partial thromboplastin time test result or by a depressed platelet count. This review summarizes published studies on the extent to which mild to moderate abnormalities of common laboratory tests of hemostasis reflect an increased risk of bleeding at the time of invasive bedside procedures, including central line placement, liver biopsy, paracentesis, thoracentesis, gastrointestinal endoscopy and biopsy, renal biopsy, bronchoscopy and transbronchial lung biopsy, and lumbar punctures. The poor predictive value of the PT (INR) is understandable in the context of the design and performance of this assay. Whether mild to moderate abnormalities of commonly used laboratory tests have any clinically relevant predictive value as appropriate triggers for prophylactic transfusions before invasive procedures should be investigated in formal randomized clinical trials.