A theoretical model was developed to describe a dynamic process involving an enzymatic reaction and diffusion of reactants and product inside glucose sensitive composite membrane. The composite membrane consisted of nanoparticles of a weakly acidic polymer, glucose oxidase and catalase embedded in a hydrophobic polymer. Time- and position-dependent diffusivity of involved species was considered in the model. Donnan equilibrium was used to find concentrations of buffer ions inside the membrane. The profiles of pH, species concentrations, volume fraction of swollen gel, polymer and water-filled space, as well as solute diffusivity inside the membrane were predicted by the model as a function of step changes of glucose concentration in the external solution. The effects of design parameters and environmental conditions on the glucose responsiveness and insulin permeability of the membrane were studied. The results indicated that the glucose responsiveness of the membrane could be amplified by optimizing the loading of glucose oxidase in the membrane. The glucose sensitivity of the membrane in a phosphate buffer was found to be higher than that in a citrate buffer for a studied membrane formulation. The presented model is a useful numerical tool for the design of glucose-responsive composite membranes for closed-loop insulin delivery. (C) 2009 Elsevier B.V. All rights reserved.