We compared the effects of freezing serum on the determination of high-density lipoprotein (HDL) subfractions by two dual-precipitation methods, heparin and manganese chloride/dextran sulfate (HM/DS) (Gidez et al., J Lipid Res 1982;23:1206-23) and DS/DS (Warnick et al., Clin Chem 1982;28:1574), and by ultracentrifugation. Storing serum for 1 month at -70 degrees C resulted in reduced HDL3-cholesterol by ultracentrifugation and reduced total and HDL3-cholesterol by the DS/DS method. There was no change in either total HDL-cholesterol or HDL3-cholesterol with the HM/DS method. Additional studies involving only HM/DS indicated that total HDL-cholesterol in serum stored at 4 degrees C begins to decline after 3 days (-3.1 +/- 3.5%, P < 0.1). HDL was stable at -20 degrees C for 2 weeks but both total and HDL3-cholesterol decreased significantly after 1 month. Storage of serum at -70 degrees C resulted in no changes for 1 year; however, at 18 months, HDL3-cholesterol was reduced 13% (P = 0.002). We conclude that HDL subfractions can be determined accurately in serum as well as in plasma after storage at -70 degrees C for up to 1 year.