Rat brain mitochondrial hexokinase (ATP: d-hexose 6-phosphotransferase, EC 2.7.1.1) was solubilized by treatment of the mitochondria with glucose 6-phosphate and partly purified. The solubilized enzyme was compared with the cytosolic enzyme fraction. The solubilized and cytosolic enzymes were also compared with the enzyme bound to the mitochondrial membrane. The following observations were made. 1. There is no difference in electrophoretic mobility on cellulose-acetate between the cytosolic and the solubilized enzyme. Both fractions are hexokinase isoenzyme I. 2. There is no difference in kinetic parameters between the cytosolic or solubilized enzymes (P < 0.001). For the cytosolic enzyme K_m for glucose was 0.067 mM (S.E. = 0.024, n = 7); K_m for MgATP²⁻ was 0.42 mM (S.E. = 0.13, n = 7) and K_i,app for glucose 1,6-diphosphate was 0.084 mM (S.E. = 0.011, n = 5). For the solubilized enzyme K_m for glucose was 0.071 mM (S.E. = 0.021, n = 6); K_m for MgATP²⁻ was 0.38 mM (S.E. = 0.11, n = 6) and K_i,app for glucose 1,6-diphosphate was 0.075 mM (S.E. = 0.010, n = 5). However when bound to the mitochondrial membrane, the enzyme has higher affinities for its substrates and a lower affinity for the inhibitor glucose 1,6-diphosphate. For the mitochondrial fraction K_m for glucose was 0.045 m (S.E. = 0.013, n = 7); K_m for MgATP²⁻ was 0.13 mM (S.E. = 0.02, n = 7) and K_i,app for glucose 1,6-diphosphate was 0.33 mM (S.E. = 0.03, n = 5). 3. The cytosolic and solubilized enzyme could be (re)-bound to depleted mitochondria to the same extent and with the same affinity. Limited proteolysis fully destroyed the enzyme's ability to bind to depleted mitochondria. 4. Our data support the hypothesis that soluble- and solubilizable enzyme from rat brain are one and the same enzyme, and that there is a simple equilibrium between the enzyme in these two pools.