Model-based techniques find a variety of applications in chemical engineering. While many components of the chemical industry have rapidly exploited model-based methods for improved operation, the pulp and paper industry has lagged behind. In this work, a detailed control study of the fiber line of a pulp mill process is presented. Based on an actual mill, the process model consists of a system of approximately 5000 states to capture the dynamics of the main units. The model includes the following major unit operations: pulp digester, oxygen reactor, bleach towers, washers, and storage tanks. Heuristic methods of plantwide control and relative gain array analysis were used to determine the primary control variables as well as input-output pairings for decentralized proportional-integral control. The performance of model predictive control and decentralized single-input single-output control were compared through closed-loop simulations for disturbance rejection and setpoint tracking.