There have been a number of attempts since the inception of the glass electrode to replace the internal filling solution with a solid state form of contact. These have included both metallic and non-metallic materials. In the present work such contacts are reviewed, the nature of the materials, and conduction mechanisms between the contact and glass are discussed. Noble metal contacts alone are not expected to give a reproducible performance because of the lack of a common charge carrier which would allow conduction across the interface. However insertion of a non-metallic transition material which contains charge carriers common to both metal and glass may lead to more predictable performance. The materials, sodium tungsten bronze, iron-aluminium-sodium silicate glass, and various silver salts were interposed between a metal conductor and ion sensitive glass. Potentiometric measurements were made to determine sensitivity and time dependence of the output potential. These revealed that both the sodium tungsten bronze and iron-aluminium-sodium silicate glass gave improved performance over noble metal contacts. The silver salt contacts gave variable performance but it was shown that composite silver salts led to greater predictability of the value of the potential. Spectroscopic measurements were made to allow association of the observed concentration changes at the interface with potentiometric data and contribute to elucidation of reaction mechanisms. It was possible to relate the observed potential changes for silver halides with the increase in silver metal content in the halide layer. Interfacial conduction mechanisms remain unclear but charge transfer is most likely to occur through ionic conduction of sodium or silver ions across the interface. Sodium containing contacts, particularly iron-aluminium-sodium silicate glass, appeared to give a more stable output potential than silver containing contacts. Fabrication of an all-solid-state glass electrode from the former class of materials may be both simpler and more likely to achieve a reliable sensor than from the latter.