Abstract: An approach integrating fuzzy comprehensive evaluation with neural network is presented to study the cutting performance of sugarcane harvester constrained by multilateral criteria of requirements. The fuzzy comprehensive evaluation method is used to study the proportion of ragged root of sugarcane,and the evaluation results are taken as training samples for neural network.Then the non-linear mapping ability of neural network is utilized to assess and forecast the cutting performance.Thus,the tedious and complicated computing process of fuzzy comprehensive evaluation is reduced and the effect of training samples error on forecast performance is lessened.The solving efficiency and self-learning process of the evaluation model is improved as well.