A detailed mathematical model was developed to describe the operation of free-radical solution styrene tubular polymerization reactors. Experiments were carried out and good agreement between steady-state model and experimental results was reached. It was also shown that the detailed model might be successfully replaced by a much simpler model, which may be implemented on-line for control purposes. Additional dynamic experimental results showed that jacket temperature perturbations may lead to oscillatory dynamic responses and that internal velocity profiles can be very distorted even at low polymer concentrations. A model predictive control, with a hybrid neural network internal model, was then implemented both theoretically and experimentally, allowing the successful control of monomer conversion.