To reduce the man-made interference in weighting the climate change vulnerability index,a weighting approach based on multi back propagation neural networks' (BPNN) global sensitivity analysis (SA) was presented through a case study of climate change assessment in hygienic field. The results show that, the BPNN fits well with a great generalization ability, each BPNNs SA result has uniformity, and the total sensitivity index is much higher than the first order sensitivity index. The fact shows that,the sensitivity analysis of the great potential influence on the climate factor presented in this paper has high stability, it can effectively identify the direct and indirect influence of the climate factors,and thus can provide a reference to the determination of the index weight.