The interaction among variables is the main problem when controlling a multivariable system, since an input signal can affect several outputs and, in turn, an output can be affected by several input signals, which is aggravated as the number of input and output signals of the system to be controlled increases. The control of a direct decoupling approach based 3x3 multivariable system is presented in this paper. At first, the direct decoupling approach allowing to eliminate or minimize the interaction among the loops of the multivariable process and to obtain a set of decoupled simple input-simple output systems is presented. After that, Proportional-Integral-Derivative (PID) controllers are designed for the set of decoupled systems by using the Internal Model Control (IMC) control strategy, and the influence of the lambda parameter in each of the outputs is analyzed. The operation of the centralized controller is then compared with a decentralized controller in presence of external disturbances. Results show that the centralized controller performs better than the decentralized controller as lower error index values and output variations were obtained.