It was found that in the octameric form of mitochondrial creatine kinase (Mr = 340 kD), only 52% of active centers bind Mg-ADP into a E-Mg-ADP-creatine complex with the dissociation constant, K(Cr)ADP, of 0.105 mM, which is close to the Km value for the enzyme (0.072 mM). In the dimeric form of cytoplasmic creatine kinase (Mr = 82 kD), 100% of active centers bind Mg--ADP; the K(Cr)ADP value (0.11 mM) is close to the Km value for the given enzyme preparation (0.083 mM). All active centers of rabbit muscle cytoplasmic creatine kinase were shown to form an analog of the transition state complex (ATSC) - E-Mg-ADP-NO3- -creatine. The constant for Mg-ADP dissociation from ATSC is identical for all centers of cytoplasmic creatine kinase and equals to 6.0 microM. The curves for ATSC saturation with Mg-ADP in the presence of iodacetamide for mitochondrial creatine kinase were constructed and computer analyzed. It was shown that in the octameric form of the enzyme only 54 +/- 13% of subunits can form ATSC. The constant for Mg-ADP dissociation from ATSC, KATSCADP is equal to 1.9 +/- 0.8 microM. It was concluded that 50% of subunits of the octameric form of mitochondrial creatine kinase are not involved in the catalytic act due to masking of their active centres and their inability to form transition state complexes. A model of regulation of cell supply with high energy compounds, e.g., ATP, creatine phosphate, via association-dissociation of mitochondrial creatine kinase oligomers is proposed.