We have developed an assay for the quantification of glycosylphosphatidylinositol (GPI)-anchored glycoconjugates. The method is based on nitrous acid deamination and sodium borodeuteride reduction of the glucosamine residue, common to all GPI structures, to yield [1-H-2]-2,5-anhydromannitol. Following acid methanolysis and trimethylsilyl derivatization, detection is by selected ion monitoring gas chromatography-mass spectrometry. The unnatural inositol isomer scyllo-inositol is used as an internal standard and the [1-H-2]-2,5-anhydromannitol trimethylsilyl derivative is detected by following a characteristic electron-impact fragment ion at m/z 273. This method is more selective for GPIs than assays based on measuring myo-inositol content, which are often confounded by contaminating inositol-phospholipids. We show that the method can be applied to measure total GPI content in crude total lipid extracts and even in whole trypanosome ghosts. The method was applied to whole cell lysates of wild-type, GPI-deficient, and glycosaminoglycan-deficient CHO cells. The data revealed that proteoglycans did not interfere with total glucosamine estimates but that there are background levels of non-GPI and nonproteoglycan glucosamine-containing material in CHO cells.