Samples from the reduced magnesium titanate spinel system ((Mg1+yTi2-yO4)-O-x+) have been examined using selected area electron diffraction (SAED) and high resolution electron microscopy (HREM) imaging. Unusual microstructural features are observed in samples in the region of the system (i.e., where x, the average Ti valency, lies between 3.26 observed in some crystals may represent the same feature viewed perpendicular to a lamellar boundary. These observations are consistent with exsolution of a single high temperature phase to two co-existing spinels of slightly different compositions. The observation of coherent lamellar boundaries and streaking of electron diffraction spots are consistent with spinodal decomposition as the mechanism of exsolution. Fine-scale lamellar structures parallel to {111} are also observed, which are texturally distinct from those described above. Streaking of electron diffraction spots parallel to < 110 > may be associated with these features. These lamellae probably represent intimate spinel-spinelloid intergrowths, rather than ordering of cation vacancies or another process. These microstructural features appear to relate to the critical resistivity transition, as they have only been observed in samples close to the compositions which have displayed such electrical behaviour. Similar studies on samples away from the region of interest fail to show these textures. A relationship between the solvus and the zero electrical resistance behaviour is inferred. In the simplest case, the zero electrical resistance material may represent a metastable homogeneous high temperature spinel from above the solvus; alternatively, the electrical behaviour may derive from the strained interlamellar boundaries. Fine-scale linear features parallel to {111}, often associated with the interlamellar boundaries, are interpreted as fine spinelloid domains; the electrical behaviour may also relate in some way to these features. It should be noted that the presence of no single microstructural feature correlates directly with the occurrence of this behaviour, nor does the loss of any feature correlate with degradation; however, it is also clear that the microstructure is continually developing as the sample ages at room temperature. Different degradation rates in air and vacuum are explained in terms of the continuing exsolution being slowed by the development of spontaneous strain at interlamellar boundaries. Chemical attack by air releases interlamellar strain and allows exsolution to progress at full speed.