The binding of nitrite and chloride in cement paste occurs mainly as a consequence of reaction with the AFm phase. The reaction has been modeled with the aid of new thermodynamic data on the stability of nitrite-AFm and by focused experiments performed on key compositions. The success of nitrite as a corrosion inhibitor for the protection of embedded steel arises from the "smart'' behavior of the AFm phase; it normally stores and sequesters nitrite in preference to sulfate, carbonate, and hydroxyl ions so that the nitrite concentrations of pore fluid are low. However, if chloride ingress occurs in service, the AFm undergoes ion exchange, gaining chloride and forming Friedel's salt, while releasing soluble nitrite ions to the pore fluid. As a result, the aqueous ratio of [NO(2)(-)]/[Cl(-)] is maintained and remains within the passivation range for steel, even though chloride increases. The exchanges are automatic and rapid, equilibrium being reached within days at similar to 25 degrees C. Thus, theory and experience can be reconciled with respect to the action of nitrite as an inhibitor of chloride-induced corrosion and the impacts of changing conditions, e. g., adding calcium carbonate or carbonation on the [NO(2)(-)]/[Cl(-)] ratio can be predicted.