Background: There is need for a reference method for transferrin glycoforms in serum to which routine immunologic methods for the alcohol marker carbohydratedeficient transferrin (CDT) can be traceable. We describe an improved HPLC method for transferrin glycoforms. Methods: Transferrin was iron-saturated by mixing the serum with ferric nitrilotriacetic acid, and lipoproteins were precipitated with dextran sulfate and calcium chloride. Separation of glycoforms was performed on a SOURCE 15Q anion-exchange column using salt gradient elution. Quantification relied on selective absorbance of the iron-transferrin complex at 470 nm. The relative amount of each glycoform was calculated as a percentage of the area under the curve, using baseline integration. Results: The HPLC system provided reproducible separation and quantification of the asialo-, monosialo-, disialo-, trisialo-, tetrasialo-, pentasialo-, and hexasialotransferrin glycoforms. Most importantly, disialo-and trisialotransferrin were almost baseline separated. The intra-and interassay CV for disialotransferrin were <5%. Serum and the pretreated samples were stable for at least 2 days at 22 or 4°C. Sera from 132 healthy controls contained [mean (SD)] 1.16 (0.25)% disialotransferrin, 4.77 (1.36)% trisialotransferrin, 80.18 (2.01)% tetrasialotransferrin, and 13.88 (1.69)% pentasialo-؉ hexasialotransferrin. In some cases of a high (>6%) trisialotransferrin, monosialotransferrin was detected at <0.25%. Asialotransferrin was not detected in control sera, but was detected in 57% of chronic heavy drinkers and in 62% of sera with >2% disialotransferrin.