The present work is aimed at the investigation of the photo-Fenton technology with regard to the remediation of diluted aqueous emulsions containing an aminosilicone polymer, in a bench-scale photochemical reactor. The experimental results show a strong interaction between temperature, light, Fe(II) and H2O2 concentrations on the degradation process, which generates substances that might be readily biodegradable and/or a solid phase that is easily separated by simple mechanical operations. The neural network technique is an effective, simple approach to successfully modeling the photo-Fenton degradation system, in which thermal and photochemical reactions and related phenomena (such as solid precipitation) take place. The model might therefore be useful in process optimization, as well as in the design and scaleup of photochemical reactors for industrial application.