Abstract: For systems with input nonlinearities, including input saturation and Hammerstein nonlinearity, the Two Step Generalized Predictive Control (TSGPC) scheme is adopted. Firstly, the input nonlinearities are not considered and linear GPC is applied to obtain the desired intermediate variable. Then, the Hammerstein nonlinearity is dealt with by solving equation and the saturation constraint is satisfied by desaturation. This paper transforms TSGPC into state space representation, and studies its domain of attraction. It transforms the solving of the domain of attraction into an iterative optimization problem and gives algorithm for this solution. It also gives the tuning algorithm for achieving desired domain of attraction. The theoretical result is validated by simulation.