Enzymatic glucose sensors are based on the amperometric detection of an oxidable species generated during the oxidation of glucose by glucose oxidase. This measurement usually requires a working electrode (anode), an auxiliary electrode (cathode), and a reference electrode, the function of the latter being to keep constant the working potential of the anode, which is responsible for current generation. However, in the needle-type glucose sensors proposed so far, the reference electrode is missing, and its function is performed by the auxiliary electrode. We investigated, in vitro and in vivo in rats, the ability of several cathode-needle materials to behave as a reference electrode in two-electrode glucose sensors, i.e., to present a stable auxiliary electrode potential. In vitro, when glucose concentration was raised from 0 to 30 mM, the auxiliary potential of both gold- and silver-coated sensors presented a cathodic drift, whereas that of silver/silver chloride-coated sensors remained stable. In vivo, during insulin-induced hypoglycemia (5.9-2.4 mM), the auxiliary potentials of all sensors remained stable, whereas during glucose infusion (mean blood glucose concentration 11.2 mM), the auxiliary potentials of both gold- and silver-coated sensors presented an anodic drift, whereas those of silver/silver chloride-coated sensors remained stable. We also indirectly quantified the changes in sensor response induced by variations in the working potential in vitro and in vivo, simulating those that might be produced by a drift in the auxiliary potential. Such changes in the working potential could bring about a 30% unspecific variation in sensor response. We conclude that improvements in sensor analytical characteristics should be obtained with silver/silver-chloride-coated cathodes.