A new uncertainty vector adjustment (UVA) was developed to compensate for the adverse effects of modelling errors and, thereby, to improve the robustness of two nonlinear control strategies : input output linearization; and Su-Hunt-Meyer transformation. The UVA provides an appropriate adjustment to the transformation relationships obtained from a nonlinear control strategy according to the overall effect of modelling errors. The effectiveness of the approach was verified by introducing parametric errors into the simulation of an evaporative stage of the liquor burning unit of the Bayer process for the production of alumina. Significant improvements in responses were observed based on the integral of the time-weighted absolute error performance index.