The discovery that approximately 20% of natural gas is microbial in origin has elevated interest in microbially enhanced coalbed methane (MECoM). However, a rational approach to exploit this calls for the development of reservoir scale models that include the effect of microbial activity. To address this, we have developed a multiscale, multiphase, multicomponent reactive -transport model for the production of microbially enhanced coalbed methane (MECoM) that includes microbial kinetics. The model is used to evaluate field scale strategies for commercial MECoM production. Optimization studies are also conducted over a range of compositions of the injected nutrient and injector bottomhole pressures. In order to account for the effect of uncertainty in the model parameters, mean -variance robust optimization is performed, allowing a trade-off between performance and robustness. Proxy modeling is performed in a multivariate polynomial chaos expansion framework to evaluate the cost functions involved in the robust optimization and sparse expansions are constructed in order to deal with issues related to high dimensionality. The optimization strategy is tested for different trade-offs between robustness and performance. It is observed that for the given case, the location of robust optimal points does not vary unless only robustness is included in the objective function, and nominal performance is not. (C) 2016 Elsevier Ltd. All rights reserved.