Abstract: Mycobacterium tuberculosis(Mtb) is a pathogen of tuberculosis, and 1-deoxy-D-xylulose-5-phosphate reductoisomerase (DXR) is a key rate-limiting enzyme in the metabolism of Mtb, so it is thought to be a potential antibacterial target. This paper collects 35 fosmidomycin derivatives with in vitro inhibitory activity against Mycobacterium tuberculosis1-deoxy-D-xylulose-5-phosphate reductoisomerase (mtDXR), conducts a three dimensional quantitative structure-activity relationship between them by using comparative molecular field analysis (CoMFA) and comparative molecular similarity indices analysis (CoMSIA), and establishes corresponding models. The results show that, for CoMFA model, the best principal component value n is 7, the cross validation coefficient q2 is 0.601,and the correlation coefficient r2 is 0.979. For CoMSIA model, the best principalcomponent value n is 8, the cross validation coefficient q2 is 0.609, and the correlation coefficient r2 is 0.983. The data show that both CoMFA and CoMSIA models have good predictability. Meanwhile, the molecular docking method is introduced to further investigate the non-bonding interaction between these small molecule inhibitors of fosmidomycin and amino acid residues on target active sites.Combined with contour maps of 3D-QSAR, the modification areas of molecular structure of these compounds are determined which can be processed and optimized.On the basis of these work, this paper designs 14 new fosmidomycin derivatives, predicts their activity, and obtains a new compound 27m with higher predictive activity. The conclusions derived from this paper provide a theoretical basis and important reference for the further development of new mtDXR inhibitors.