This work studies the experimental application of the globally linearizing control (GLC) method to a batch polymerization reactor. The nonlinear controller is implemented on a microcomputer to start up the reactor and then track a precalculated optimal temperature profile. The reactor temperature is controlled by manipulating two coordinated inputs: power to an electrical heat and cooling water flow rate. A reduced-order observer is used to estimate the concentration of initiator and monomer. Systematic tuning guidelines are proposed for the nonlinear control method. The experimental results show the excellent servo and regulatory performance of the nonlinear controller in the presence of modeling and observer initialization errors and active manipulated input constraints. Furthermore, in comparison to a conventional PID controller, the performance of the nonlinear controller is significantly superior, and its tuning is much easier.