Within industrial process development, powerful screening techniques are required to select the optimal biocatalyst regarding such process characteristics as cost effectiveness, turnover number or space time yield. Conventional measurement of the initial enzyme activity, which is the established high throughput screening technique, disregards the long-term stability of an enzyme. A new model based technique called “enzyme test bench” was recently presented before by our group which addresses this issue. It combines the high throughput screening approach with an extensive enzyme characterization, focusing especially on the long-term stability. The technique is based on modeling enzyme activation and deactivation as temperature dependent reactions in accordance with the Arrhenius law. Controlling these reactions by tailor made temperature profiles, the slow long-term deactivation effects are accelerated and characterizing models are parameterized. Thus, the process properties of an enzyme can be predicted and included into the screening procedure. Moreover, the optimum process temperature as function of the envisaged operation time can be found by these means. In this work, the technique is extended to the important class of oxygen consuming reactions. For this aim, a suitable assay and a defined oxygen supply were established. This extended technique was applied to characterize and to optimize a complex, multi-stage laccase-mediator system (LMS). For the variation and optimization of the enzyme to mediator to substrate ratio, experiments in microtiter plates were performed. Predictions from this high throughput characterization were compared to long-term experiments in a RAMOS device (Respiration Activity Monitoring System), a technique for on-line monitoring of the oxygen transfer rate in shake flasks. Within the limits of the model validity, the enzyme test bench predictions are in good agreement with the long-term experiments. Biotechnol. Bioeng. 2014;111: 244–253. © 2013 Wiley Periodicals, Inc.