In this paper we take photobacterium phosphoreum(T_3)as the experimental bacteria, and determine the half-inhibitory concentration(-lgEC_(50))against the photobacterium phosphoreum of 16 halogenated benzenes.Using B3LYP method of DFT in the Gaussian 03 program,we obtain the structural and thermodynamic descriptors of 16 halogenated benzenes by fully-optimized calculation at the 6-311G** level.Taking the structural and thermodynamic descriptors as theoretical descriptors,the 2D QSAR model(R~2=0.983)was established,which can be utilized to predict-1gEC_(50)of halogenated benzene according to the corrected linear solvation energy theory based on the experimental data of-1gEC_(50).In addition,the relationship between the toxicity and 3D spatial structure of the compound is studied by comparing the molecular similarity index analysis (CoMSIA)of 3D-QSAR method.By cross validation,the correlation coefficient q~2 of CoMSIA model is 0.687,and the conventional correlation coefficient R~2=0.958.The model is stable and reliable with great predictive ability.The 3D-QSAR model shows that the toxicity of halogenated benzene compound is mainly affected by the characteristics of hydrophobic field of the substituted halogens. In this paper we take photobacterium phosphoreum (T_3) as the experimental bacteria, and determine the half-inhibitory concentration (-lgEC_ (50)) against the photobacterium phosphoreum of 16 halogenated benzenes. Using B3LYP method of DFT in the Gaussian 03 program, we obtain the structural and thermodynamic descriptors of 16 halogenated benzenes by fully-optimized calculation at the 6-311G ** level. The structural and thermodynamic descriptors as theoretical descriptors, the 2D QSAR model (R ~ 2 = 0.983) was established, which can be utilized to predict-1gEC_ (50) of halogenated benzene according to the corrected linear solvation energy theory based on the experimental data of-1gEC_ (50) .In addition, the relationship between the toxicity and 3D spatial structure of the compound is studied by comparing the molecular similarity index analysis (CoMSIA) of 3D-QSAR method. By cross validation, the correlation coefficient q ~ 2 of CoMSIA model is 0.687, and the conventional correlation coefficient R ~ 2 = 0.958.T The model is stable and reliable with great predictive ability. The 3D-QSAR model shows that the toxicity of halogenated benzene compounds is primarily affected by the characteristics of hydrophobic fields of the substituted halogens.