Since the seminal work of Smith, predictor structures have been used to control processes with dead-time. Predictors allow the control of this type of process with a delay-free nominal model, which simplifies the control design. In this paper, a nonlinear filtered Smith predictor (NLFSP) structure is proposed for systems with input nonlinearities. It is shown that the NLFSP maintains the characteristics of the linear filtered Smith predictor and that with appropriate tuning, it can increase the robustness of the closedloop system or improve the disturbance rejection response. The NLFSP is applied to a simulated CSTR process to demonstrate these characteristics. (C) 2015 Elsevier Ltd. All rights reserved.