Abstract: A series of novel imine heterocyclic compounds containing thiazole and triazole ring were classified by Var.Jarvis-Patrick clustering method coupled with principal component analysis (PCA) and then the quantitative structure-activity relationships (QSAR) were investigated using genetic function approximation (GFA) and molecular field analysis (MFA),respectively.All models were cross-validated with CV-r2 over 0.910 and used to predict the compounds with reported activity.The results demonstrate that all models are significant with good stability and predictability.The 2D-results indicated that molecular thermodynamic properties (atom type AlogP descriptors),steric states (Jurs and Shadow parameters) and electrotopological state indice (S_aaCH) contribute to the activity.The steric function is mostly involved in activity demonstrated by 3D-study.A series of imine compounds were designed according to the theory of drug design and their activity was predicted and reciprocally validated using optimized 2/3D-QSAR models.Six compounds with potential high activity were finally screened.In conclusion,the results in the research are valuable for designing novel effective imine fungicides.