Structure-based algorithm for fingerprint recognition fits well into the need of general solid-state captures that have limited wafer area. It makes use of the abundant structure information of fingerprint image, and moreover, its Gabor feature vectors have equal length good for quickly matching. We designed optimal Gabor filters and corresponding fingerprint representations for the fingerprint recognition . In order to improve the accuracy, we proposed a matching algorithm using an Adaptive Neuro-Fuzzy Inference Systems (ANFIS) network, which is trained to identify the Gabor features of fingerprints. The subtractive clustering algorithm and the least-squares estimator are used to identify the fuzzy inference system. The training process is accomplished by using the hybrid-learning algorithm. In this paper, the construction of ANFIS is described in detail. The experimental demonstration is reported, which proves that this matching algorithm could achieve a high accuracy. The comparison with the best algorithm in FVC2000 is presented, the result analysis and future work are given in the end. We developed a new application field of ANFIS, and this method can also be used for other pattern recognition applications.