Comparative quantitative-structure-toxicity-relationship models developed using various quantum chemical descriptors calculated at semiempirical (AM1, PM3), Hartree–Fock and density functional theory (DFT) (B3LYP) methods, are compared for predicting the toxicity (pIGC50) of 69 aliphatic compounds toward Vibrio fischeri. The multiple linear regression equations, which can assist in modeling toxicity of these compounds, are developed using molecular descriptor, log P (1-octanol/water partition coefficient) in conjunction with two quantum chemical descriptors, electrophilicity index (ω) and energy of the lowest unoccupied molecular orbital (ELUMO). The calculated toxicity values (PIGC50) based on the model equations are compared with the experimental data. In general, the stepwise model regression equations based on quantum chemical descriptor ω in combination with molecular descriptor log P, calculated at DFT(B3LYP) with 6-311G (d,p) and DGDVZP basis sets show better agreement with the experimental data. The electronic origin of the toxicity dependence on descriptors like log P, ω and ELUMO is traced to the cumulative contribution of the lipophilicity and electrophilicity of the chemical moieties to their toxic behavior. © 2012 Wiley Periodicals, Inc.