Nonlinear modelling approaches such as neural networks, fuzzy models and multiple model networks have been proposed for model based control, to improve the poor transient response of adaptive control techniques. The quality of control is known to be strongly related to the accuracy of the model which represents the process. A Bayesian Gaussian process (GP) approach provides an analytic prediction of the model uncertainty, which makes the GP model an ideal candidate for model based control strategies. This article extends the use of the GP model for nonlinear internal model control. The invertibility of the GP model is discussed and the use of predicted variance is illustrated on a simulated example.