We present experimental and theoretical evidence that the high field conduction mechanism for buried oxide prepared by the technique of separation by implanted oxygen (SIMOX) is due to Fowler-Nordheim (F-N) injection aided by field intensification at interfacial asperities. The asperities are believed to be the pyramidal protusions thought to be stacking faults, observed in electron microscopy (XTEM) studies. The standard F-N theory is modified to include field intensification (k(FI)) and reduced area (k(A)) factors to account for the presence of these defects. Fitting of the model to experimental current-voltage data yields a value for k(A) which predicts an average asperity separation in agreement with XTEM and defect etching studies. This work therefore challenges an existing theory which considers that the high field conduction involves tunnelling between Si clusters arising from the excess Si known to exist in SIMOX. We present arguments based on experimental observations relating to constant voltage stressing experiments, that our model can explain the observed behavior and, in addition, the anomalies relating to the excess Si tunneling model.