Second-order rate constants (k(O3)) for the reaction of ozone with micropollutants are essential parameters for the assessment of micropollutant elimination efficiency during ozonation in water and wastewater treatment. Prediction models for k(O3) were developed for aromatic compounds, olefins, and amines by quantum chemical molecular orbital calculations employing ab initio Hartree Fock (HF) and density functional theory (B3LYP) methods. The k(O3) values for aromatic compounds correlated well with the energy of a delocalized molecular orbital first appearing on an aromatic ring (i.e., the highest occupied molecular orbital (HOMO) or HOMO-n (n >= 0) when the HOMO is not located on the aromatic ring); the number of compounds tested (N) was 112, and the correlation coefficient (R-2) values were 0.82-1.00. The k(O3) values for olefins and amines correlated well with the energy of a localized molecular orbital (i.e., the natural bond orbital (NBO)) energy of the carbon carbon pi bond of olefins (N = 45, R-2 values of 0.82-0.85) and the NBO energy of the nitrogen lone-pair electrons of amines (N = 59, R-2 values of 0.81-0.83), respectively. Considering the performance of the k(O3) prediction model and the computational costs, the HF/6-31G method is recommended for all aromatic groups and olefins investigated herein, whereas the HE/MIDI!, HF/6-31G*, or HF/6-311++G** methods are recommended for amines. Based on their mean absolute errors, the above models could predict k(O3) within a factor of 4, on average, relative to the experimentally determined values. Overall, good correlations were also observed (R-2 values of 0.77-0.96) between k(O3) predictions by quantum molecular orbital descriptors in this study and by the Hammett (sigma) and Taft (sigma*) constants from previously developed quantitative structure activity relationship (QSAR) models. Hence, the quantum molecular orbital descriptors are an alternative to sigma and sigma*-values in QSAR applications and can also be utilized to estimate unknown sigma or sigma*-values.