We describe a model of mitochondrial regulation in vivo which takes account of spatial diffusion of high-energy (ATP and phosphocreatine) and low-energy metabolites (ADP and creatine), their interconversion by creatine kinase (which is not assumed to be at equilibrium), and possible functional 'coupling' between the components of creatine kinase associated with the mitochondrial adenine nucleotide translocase and the myofibrillar ATPase. At high creatine kinase activity, the degree of functional coupling at either the mitochondrial or ATPase end has little effect on relationships between oxidative ATP synthesis rate and spatially-averaged metabolite concentrations. However, lowering the creatine kinase activity raises the mean steady state ADP and creatine concentrations, to a degree which depends on the degree of coupling. At high creatine kinase activity, the fraction of flow carried by ATP is small. Lowering the creatine kinase activity raises this fraction, especially when there is little functional coupling. All metabolites show small spatial gradients, more so at low cytosolic creatine kinase activity, and unless there is near-complete coupling, so does net creatine kinase flux. During workjump transitions, spatial-average responses exhibit near-exponential kinetics as expected, while concentration changes start at the ATPase end and propagate towards the mitochondrion, damped in time and space.