BACKGROUND Accurate measurement of blood glucose concentrations is essential for defining diabetes, and the minimization of ex vivo glycolysis has been recommended. Recent guidelines advocate two kinds of methods for sample collection and processing: either the sodium fluoride (NaF) method or immediate refrigeration using a serum separation tube (SST). We investigated the difference between the two methods in measuring subsequent glucose concentrations using blood specimens from participants recruited for the fourth Korean National Health and Nutrition Examination Survey. METHODS Paired venous blood samples were collected in an SST and a NaF tube from 1,103 men and women. SST serum was separated within 30 min, including standing for 15 min, and then refrigerated. The NaF samples were refrigerated, but not separated until immediately before analysis. We compared the blood glucose concentrations between the SST (SST glucose) and NaF (NaF glucose) methods. RESULTS The mean SST glucose was significantly higher than NaF glucose (99.0 mg/dL vs 96.5 mg/dL, P<0.05). NaF glucose showed a negative mean bias of 2.6 mg/dL vs SST glucose but showed high correlation (R=0.9899). There was no significant correlation between the bias of blood glucose concentrations by two methods and the storage time of NaF glucose. CONCLUSIONS The negative bias associated with the use of NaF tubes may significantly affect the prevalence of diabetes. Serum separation and refrigeration within 30 min after venous sampling is recommended over NaF method, not only to minimize the preanalytical impact on detecting diabetes but also to reduce sample volume and number of tubes.