Model Predictive Control (MPC) is a useful tool when controlling processes that handle a large number of input and output variables. This study presents a comparison of different MPC strategies when they are subjected to control process variables directly. The strategies studied are IMC, GPC, MPC-D, MPC-DR, and DMC. Evaluation of the performance of the controlled loop was performed with the filtering and correlation analysis algorithm (FCOR). The methodology proposed is validated in a Continuous Stirred-Tank Reactor (CSTR) case study. Discrete predictive control demonstrated the best results in this study.