A mathematical model is developed of the compartmentalized sialylation of N-linked oligosaccharides in order to understand and predict the outcome of sialylation reactions. A set of assumptions are presented, including Michaelis-Menten-type dependency of reaction rate on the concentration of the glycoprotein substrate. The resulting model predicts the heterogeneous outcome of a posttranslational oligosaccharide biosynthesis step, a critical aspect that is not accounted for in the modeling of the cotranslational attachment of oligosaccharides to glycosylation sites (Shelikoff et al., Biotech. Bioeng., 50, 73-90, 1996) or general models of the secretion process (Noe and Delenick, J. Cell Sci., 92, 449-459, 1989). In the steady-state for the likely case where the concentration of substrate is much less than the Km of the sialyltransferase, the model predicts that the extent of sialylation, x, will depend upon the enzyme concentration, enzyme kinetic parameters and substrate residence time in the reaction compartment. The value of x predicted by the model using available literature data is consistent with the values of x that have been recently determined for the glycoproteins CD4 (Spellman et al., Biochemistry, 30, 2395-2406, 1991) and t-PA (Spellman et al., J. Biol. Chem., 264, 14100-14111, 1989) secreted by Chinese hamster ovary cells. For the unsaturated case, the model also predicts that x is independent of the concentration of secreted glycoprotein in the Golgi. The general modeling approach outlined in this article may be applicable to other glycosylation reactions and posttranslational modifications.