The removal of gas-phase styrene was studied in both one- and two-liquid phase continuous stirred tank bioreactors (CSTBs) inoculated with Sporothrix variecibatus. Experiments were carried out at different gas residence times and inlet styrene concentrations to reach inlet loading rates varying between 10 and 838 g m(-3) h(-1) in the bioreactors. The addition of 10% (v/v) silicone oil acted as a buffer for high styrene loads and improved the two-liquid phase CSTB performance 3.1-fold in comparison to the one-liquid phase CSTB, with maximum elimination capacities of 426 and 137 g m(-3)h(-1), respectively. The CSTBs performance was modeled using artificial neural networks (ANNs) with inlet concentration (g m(-3)) and unit flow (h(-1)) as the input variables. The best network topology, selected by a trial and error approach and by estimating the determination coefficient (R-2) values, was found to be 2-5-1 and 2-6-1, respectively, for the one- and two-liquid phase CSTB.