DNA testing is often considered a gold standard in clinical laboratory testing. But just how golden is the standard? In this issue of American Journal of Hematology, Emadi et al. investigate this question with regard to the two most commonly performed DNA tests in the field of coagulation: Factor V Leiden and prothrombin G20210A [1].Unlike other coagulation assays, such as protein C, protein S, or antithrombin, the presence or absence of factor V Leiden or prothrombin G20210A is constant throughout an individual's lifetime. Thus, clinicians often assume that there is no need to repeat the DNA test to confirm the findings. Is this a valid approach? In comparison, for example, protein C can be reduced by liver dysfunction, vitamin K deficiency, poor nutrition, oral anticoagulation therapy, disseminated intravascular coagulation (DIC), recent thrombosis or surgery, or patient age (protein C is lower than adults throughout childhood). In addition, some protein C activity assays may be falsely elevated by lupus anticoagulants or direct thrombin inhibitor anticoagulant therapy. Therefore, repeat testing and family studies are often needed to investigate the possibility of hereditary protein C deficiency. In contrast, DNA testing is not affected by any of these variables. Thus, is it safe to assume that the results of a DNA test truly represent the patient's genotype?Emadi and colleagues sought to answer these questions by carefully reviewing all relevant abstracts (studies comparing a factor V Leiden and/or prothrombin G20210A DNA assay to a reference method) published between 2000 and 2008, analyzing the 66 manuscripts that met their selection criteria. The reference method in most of these studies was polymerase chain reaction followed by restriction fragment length polymorphism (PCR-RFLP) or allele-specific PCR. Concordance with the reference method was 98-100% for all methods except for some studies with agreement as low as 90.2%. In most cases, the discrepancy resolved upon repeating the test. Many of the errors may have been attributable to human error rather than a failure of the assay itself. The authors conclude that the error rate is low. However, it is important to emphasize that errors do occur, for example, 98% concordance means that one in every 50 specimens is reported incorrectly. Furthermore, studies on external quality assurance (EQA, proficiency programs that assess a clinical laboratory's ability to correctly test unknown specimens) revealed that most laboratories were highly accurate with these tests, but errors included transcription errors (writing the wrong result when preparing the report), misinterpretation of the data, and in one study, a homozygous factor V Leiden specimen was misidentified by 15% of participating laboratories. In another EQA study, 51% of laboratories made at least one error over time (133 specimens were sent to 39 laboratories over 10 different periods of time), but the authors note that three of 39 laboratories were responsible for 46% of all errors.Additional EQA data f...