BACKGROUND: 4-Hydroxyphenylpyruvate dioxygenase (HPPD) herbicides control broadleaf and gramineous weeds with better crop safety for corn, sorghum and wheat. Multiple screening models in silico have been established to obtain novel lead compounds as HPPD inhibition herbicides.RESULTS: Topomer comparative molecular field analysis (CoMFA) combined with topomer search technology and Bayesian, genetic approximation functions (GFA) and multiple linear regression (MLR) models generated by calculating different descriptors were constructed for the quinazolindione derivatives of HPPD inhibitors. The coefficient of determination (r2 ) of topomer CoMFA, MLR and GFA were 0.975, 0.970 and 0.968, respectively; all the models established displayed excellent accuracy and high predictive capacity. Five compounds with potential HPPD inhibition were obtained via screening fragment library combined with the validation of the above models and molecular docking studies. After molecular dynamics (MD) validation and absorption, distribution, metabolism, excretion and toxicity (ADMET) prediction, the compound 2-(2-amino-4-(4H-1,2,4-triazol-4-yl) benzoyl)-3-hydroxycyclohex-2-en-1-one not only exhibited stable interactions with the protein but also high solubility and low toxicity, and has potential as a novel HPPD inhibition herbicide.CONCLUSION: In this study, five compounds were obtained through multiple quantitative structure-activity relationship screening. Molecular docking and MD experiments showed that the constructed approach had good screening ability for HPPD inhibitors. This work provided molecular structural information for developing novel, highly efficient and low-toxicity HPPD inhibitors. © 2023 Society of Chemical Industry.