The samples studied in this paper were derived from the iron oxide facies of iron formations (IFs) of the Quadriltero Ferrfero, south-eastern Brazil. The rich-iron oxide varieties of the IFs can be regarded as Fe-O systems because they consist predominantly of aggregates of haematite and magnetite crystals. The study was based on carefully selected samples in which magnetite crystals occur as isolated large crystals set in a matrix of haematite grains. Optical microscopy and an electron microscope equipped with an electron backscattered diffraction (EBSD) system were used to investigate the microstructural and crystallographic relationships involved in the phase transformation between magnetite and haematite. The magnetite crystals in the IF samples are partially transformed to haematite and this process is also known as martitisation. This transformation occurs along the crystallographic planes {111} of the magnetite producing a characteristic triangular pattern of interlocking lamellae of transformed haematite crystals that are oriented parallel to octahedral plane {111}. This new generation of haematite is referred to martite because of the martitisation process. EBSD analyses show that the magnetite and the new haematite crystals, martite, share crystallographic planes of highest atomic density, eg {111} and (0001), respectively, possibly due to the ccp and hcp atomic arrangement of the two crystals. Stereographic projections of these two planes therefore coincide when the maxima of their poles are plotted. However, the crystallographic orientation of the haematite crystals in the matrix shows maxima of (0001) poles that do not coincide with those of {111} planes of magnetite. Consequently, it is concluded that the direct transformation to haematite crystals retains the crystallographic orientation memory of the former magnetite crystals, possibly due to the similar manner whereby their atoms are packed together in the crystal structures. On the other hand, no such relationships are observed between magnetite and haematite crystal in matrix. This suggests that haematite grains in the matrix may have been formed by other processes rather than those involved in the iron oxide transformation observed within the magnetite grains.