This paper proposes a model-based reference tracking scheme for stable, MIMO, nonlinear processes. A Joint Unscented Kalman Filtering technique is exploited here to develop a stochastic model of the physical process via simultaneous estimation of the process states and the time-varying/uncertain parameters. Unlike the existing nonlinear model predictive controllers, the proposed scheme does not involve any dynamic optimisation process, which helps to reduce the overall complexity, computation overburden and execution time. Furthermore, the proposed methodology offers robustness to process model-mismatch and considers the effects of stochastic disturbances. A nonlinear two-tank liquid-level control problem and a nonlinear coupled level-temperature control process are studied to demonstrate the usefulness of the proposed scheme.