Despite the many advantages of biocatalysis, enzyme stability remains an issue. This work investigates the long-term kinetic stability of a water-forming NAD(P)H-oxidase (NOX2), an enzyme of potential industrial interest for NAD(P)H regeneration, in a bubble column sparged with air, which enables reasonable oxygen transfer but without drastic enzyme deactivation. Experiments have been performed with the particular goal of explaining the observed two-stage deactivation trend. We show evidence supporting the hypothesis that the first stage is related to the adsorption of enzymes to the interface, followed by a subsequent deactivation stage at the interface.The characterization of NOX deactivation in the bubble column setup, complemented by additional’quiescent’ experiments, has enabled the development of a first principles model as the first step towards a complete mechanistic model of oxidase deactivation at gas–liquid interfaces.