This paper presents a genetic algorithm (GA) based autotuning method to design a decentralized proportional-integral (PI) control system for composition control of a highly interactive and nonlinear reactive distillation column. The control relevant characteristics such as nonlinearities, interactions, and stability are analyzed for assessing the complexity of the process. The objective of GA tuning is to account the multivariable interactions and nonlinear dynamics of the process to find a unique set of parameters for the control system that is robust to all kinds of disturbances. The performance function in GA is formulated by incorporating the dynamic state information of the process derived from its model for various closed-loop disturbance conditions. The controller tuning problem of this multivariable process is resolved as an optimization problem and multiloop PI controllers are designed by exploiting the powerful global Search features of GA. An estimator is designed to provide the compositions which serve as inferential measurements to the controllers. The performance of the proposed GA-tuned decentralized control scheme is evaluated by applying it to a metathesis reactive distillation column, and the results are compared with conventionally tuned PI controllers. The results demonstrate the better regulatory and servo performance of the GA-tuned PI controllers for composition control of reactive distillation column.