Conventional experimental design techniques are available to assist in the optimization of fermentation processes, but due to the nonlinearities in the bioprocess, they are limited in their effectiveness. This problem is further complicated with recombinant systems as a result of the additional complexities of the process. This article describes a general strategy using artificial neural networks as an alternative approach to fermentation process development laboratory are presented for the neural network based procedures. (c) 1994 John Wiley & Sons, Inc.