In this study, a one-hidden-layer Artificial Neural Network (ANN) using a back-propagation structure has been trained on experimental data for the identification of styrene conversion and polymer average molecular weight produced in a bulk polymerization initiated by bifunctional peroxides. Two identification schemes, a leave-N-out and black-box with constant and adaptive step-size, are presented. Both schemes gave good predictions of monomer conversion and molecular weight averages. Neural networks for nonlinear identification of monomer conversion and average molecular weights can be used along with other approaches to improve modeling of polymerization reactions, especially when the kinetic mechanism is not well established.