In spite of much work, the formation of epitaxial CoSi2 from Ti/Co on (100) Si remains something of a mystery. It has been proposed that epitaxy occurs via the formation of an intermediate phase of CoSi with a (311) preferred orientation. In the absence of sufficient information it is impossible to validate or to invalidate the specific original claim. However, one shows that the formation of preferably oriented CoSi is not a necessary condition for the subsequent growth of epitaxial CoSi2. Careful measurements of diffraction intensities reveal the probable, temporary formation of a metastable form of CoSi2, based on a diamond cubic rather than the usual CaF2 structure. © 1995 American Institute of Physics.