A previously developed mathematical model for ethylene polymerization in a fluidized bed reactor was used to investigate the process static and dynamic behavior. The static analysis determined optimal operating conditions at which the monomer conversion can be increased to 25% per pass. However, this optimal operating. point was shown to be unstable, and thus any changes in the plant operation may lead to temperature runaway, degrading the reactor performance. In addition, the reactor temperature must be kept within a narrow range between the gas dew point and the polymer melting point. For these reasons, two control algorithms, that is, proportional-integral (PI) and nonlinear model predictive control (NLMPC), were tested for the stabilization of the process during load changes. The PI closed-loop dynamic simulations indicated that using the reactor feed temperature solely as a manipulated variable was not sufficient to stabilize the reactor temperature. However, good PI feedback performance was obtained when the feed temperature and the superficial velocity were used to control the reactor temperature and the monomer concentration. This control structure was selected arbitrarily whereas the control structure recommended by standard methods such as RGA and SVD performed poorly. On the other hand, the NLMPC out-performed the PI control in terms of minimum tuning and controller design efforts.