Teaching quality evaluation system is a multi-level and multi-target complex issue, with numerous evaluation indexes, and to take all evaluation indexes as input of neural network would cause complex structure of neural network, thus affecting the performance of teaching quality evaluation system. This paper has proposed a teaching quality evaluation method based on AHP-BPNN. Firstly, the AHP is used to sort importance of evaluation index system, and then indicators that have an important impact on evaluation results are filtered out as input of BPNN, and finally, evaluation model is established using NN. Simulation experiment is made in the case of "Data Structure" in Fuyang Normal College, and the simulation results show that, AHP-BPNN not only simplifies the structure of the neural network, but also improves the evaluation accuracy and efficiency of the teaching quality, thus it is a feasible and effective method of teaching quality evaluation.