The production of GA3 in solid-state culture was studied using systems composed of maize cob particles soaked in an amylaceous effluent, as a function of three variables-particle diameter, volume of liquid phase, and substrate concentration-with a determinant influence on the response. The empirical model obtained provides an explanation of the behavior of the system in terms of a series of interactions with underlying problems of water activity and mass transfer, that can, however, be manipulated in a simple and reproducible manner. This model also allows the suitable orientation of production improvement, the interpretation of the mechanism of substrate inhibition detected, as well as means of correcting it without reducing the initial concentration of the substrate, and leads to an 11-fold increase in the production obtained in the preliminary assays.