Whereas factors governing the interconversion of the two oligomeric forms of mitochondrial creatine kinase are relatively well known, few informations are yet available on the actual form in situ. Antibodies against purified pig and rabbit heart mitochondrial creatine kinase were obtained. The former exhibits a marked specificity for the dimer while the second reacts with both dimer and octamer. They allowed to demonstrate that no dimer can be detected in mitochondria and that CKm occurs naturally exclusively as an octamer. We present arguments that the larger part, if not the totality, of the octamer is membrane-bound rather than soluble in the intermembrane space. However, these findings do not refute the previously proposed models for the regulation of CKm activity in the mitochondrion but urge to envisage a more complex one.