A full factorial design was used to study the influence of four different variables, namely polymer concentration, carbodiimide concentration, time of reaction and blocking agent concentration, on the coupling of a serine protease into a soluble–insoluble polymer (Eudragit S‐100). All of the four factors studied have played a critical role in the protease coupling. Response surface methodology was used as an optimisation strategy to attain a conjugate with high activity yield and operational stability at 60 °C. Under optimised conditions (Eudragit, 2.5% w/v, carbodiimide, 0.2% w/v, coupling time, 1 h and blocking agent concentration, 0.05%), the conjugate activity yield was about 45% and its operational stability at 60 °C was increased by 1.7 times. After reusing the conjugate for five cycles, the remaining activity was still 72% of the initial value when compared with the native enzyme. Several tests confirmed that the enzyme was covalently crosslinked to Eudragit, which represents an improvement in the carbodiimide coupling of proteases into soluble–insoluble polymers. Copyright © 2005 Society of Chemical Industry