In this paper, a new fuzzy neural network structure called the fuzzified input neural network (FINN) is presented. The FINN structure is derived based on human knowledge in the form of fuzzy if-then rules. The initial setting of its parameters can be chosen intuitively from expert experience. These parameters are then adaptively adjusted during system operation. Comparisons between the proposed network and the conventional Mamdani fuzzy inference are described. The performance of FINN is demonstrated by using it as the controller for the single inverted pendulum plant. Some simulation results are given.