Abstract: We synthesize a stable predictive controller based on the Hammerstein model by using the nonlinear separation method. Firstly, we obtain the desired intermediate variable by using the finite-horizon stable predictive control method, and then we obtain the real control action by solving nonlinear algebraic equations. While calculating the desired intermediate variable, we consider the uncertainty in input nonlinearity and the error in solving nonlinear algebraic equations. In determining the terminal constraint set and terminal state weighting matrix for the finite-horizon optimization problem, we utilize the parameter-dependent method and introduce the reference set to optimize the terminal constraint set. The algorithm is validated with a simulation example.