AbstractBiological mineralization is a cell‐mediated process which is believed to be triggered by a “nucleating agent.” Various matrix structures, such as matrix vesicles, collagen fibrils and macromolecules, have been claimed to be the source of this substance, since these components have been found by transmission electron microscopy (TEM) of thin sections to be associated with early mineral crystals. Systematic image analysis of the relationships revealed in electron micrographs between specific matrix components and early mineral deposits has shown that unequivocal image interpretation is not possible. This is due principally to the problems posed by overprojection and truncation phenomena, since the structures being analyzed lie within the same dimensional range as thin section thickness. Various examples are illustrated and discussed. The site at which mineral crystals are initially laid down thus cannot be identified with any matrix structure using thin section TEM. Possible technical approaches to resolve this problem of image analysis are disc