Thermal oxidation of a silicon-on-insulator substrate produces evidence that an ordered SiO2 structure can exist on thermally oxidized SiO2–Si interfaces. An apparently ordered SiO2 layer was observed by a high-resolution transmission electron microscope (HRTEM) when a thin silicon layer enclosed by SiO2 was less than 3.0 nm thick. X-ray diffraction of the ultrathin Si (≪3 nm) samples showed diffractions from an ordered SiO2 phase, first-order Bragg reflection peaks with a lattice spacing of 4.1±0.15 Å, and second-order Bragg reflection peaks with 2.03±0.15 Å, in addition to the peaks from the Si substrate and the thin Si layer. Even in samples with thick Si layers enclosed by SiO2, which did not show the apparently ordered silicon oxide layer by the HRTEM, x-ray results showed a weak diffraction as if from a crystalline silicon oxide. The disappearance of the second-order Bragg reflection at higher energies indicates that the lattice structure of any crystalline SiO2 phase is far from perfect.