Abstract: A new realizing strategy of chemotaxis algoritbm for neural networks training is developed.Modeling with dynamic recurrent neural networks and nonlinear control strategy with the constraints on control action using neural networks are prese-nted.They are applied for continuous fermentor.Simulation results show that neural network prediction and control strategies are robust for variations in plant parameters and accurate with a certain degree of noise immunity.They offer the distinctive ability over more traditional methods to learn very naturally complex relationship without requiring the knowledge of the model structure,Nonlinear control strategy based on neural networks is effective for controlling of continuous fermentor which is character-ized by zero steady-state gain with respect to one manipulated input at the optimum opterating point and attendant change in sign of the steady-state gain across the optimum.