In this paper neuro-fuzzy predictive controller for nonlinear system is addressed, proposed and tested. The proposed neuro-fuzzy convolution model consists of a steadystate neuro-fuzzy model and a gain independent impulse response model. The proposed model is tested in model based predictive control of a real laboratory plant. The basic principles of predictive control algorithm for thermo-optical plant are proposed. The paper deals with theoretical and practical methodology, offering approach for intelligent neuro-fuzzy control design and its successful application.