There is much interest in the use of solid state fermentation (SSF) for the production of a range of microbial enzymes, and a number of commercial processes currently exist. However, due to the poor heat transfer characteristics of beds of solid substrate it is difficult to prevent rises in temperature of the fermentation medium during large scale SSF processes. This can have adverse effects on enzyme levels in the product, due to temperature-mediated enzyme deactivation during the fermentation. We undertook a modeling investigation to explore the likely extent of problems in commercial production processes involving SSF. The system modeled involves the production of an acid protease by the fungus Rhizopus oligosporus. At the neutral pH values which are optimal for fungal growth, this protease is unstable at the temperature optimum of growth of 37ºC, slowly losing activity over a number of hours. Protease activity in the substrate was modeled as a result of two processes: production by the fungus, with both growth-associated and non-growth associated components, and thermal deactivation. Deactivation was described as a two-step processes involving a semi-active intermediate, with each step occurring as a first-order process. The model was fitted to deactivation curves at various temperatures to determine the first-order rate constants and the relative activity of the semi-active intermediate. The first order rate constants varied with temperature according to the Arrenhius equation. The model was fitted to profiles of biomass and protease activity obtained at 37ºC in order to extract the parameters associated with enzyme production. These equations were combined with an energy balance written for a large scale SSF bioreactor, and the model was used to explore the likely performance of the bioreactor in a protease production process. The model predictions suggest that it will be difficult to optimize production. The rate of protease deactivation decreases as the temperature decreases below the optimum for growth, but low temperatures also decrease the rate of protease production due to the growth-associated production kinetics. It is obvious that accurate models of enzyme production and stability can play an important role in optimizing production processes for such "semi-stable" enzymes.