This paper proposes a hybrid process modeling and optimization formalism integrating artificial neural networks (ANNs) and genetic algorithms (GAs). The resultant ANN−GA strategy has the advantage that it allows process modeling and optimization exclusively on the basis of process input−output data. In the hybrid strategy, first an ANN-based process model is developed from the input−output process data. Next, the input space of the model representing process input variables is optimized using GAs, with a view to simultaneously maximize multiple process output variables. The GAs are stochastic optimization methods possessing certain unique advantages over the commonly used gradient-based deterministic algorithms. The efficacy of the hybrid formalism has been evaluated for modeling and optimizing the zeolite (TS-1)-catalyzed benzene hydroxylation to phenol reaction whereby several sets of optimized operating conditions have been obtained. A few optimized solutions have also been subjected to the experimenta...