Abstract Quantitative structure–property relationship/quantitative structure–activity relationship (QSPR/QSAR) models were developed for rate constants ( k) of alkylnaphthalene reactions with chlorine (Cl), hydroxyl ( OH) and nitrate (NO 3) radicals using partial least squares (PLS) regression. Quantum chemical descriptors computed by Parametric Method 3 (PM3) Hamiltonian were used as predictor variables. The cross-validated Q cum 2 values for the optimal QSPR/QSAR models of alkylnaphthalenes are 0.896, 0.728 and 0.774 for Cl, OH and NO 3 radicals, respectively. Results from this study showed that rate constants with Cl, OH and NO 3 are governed by different molecular structural descriptors. In the developed optimal QSPR/QSAR models, frontier molecular orbital energies and atomic charges are major descriptors that affect log k values. When the highest occupied molecular orbital ( E HOMO) energy, the lowest unoccupied molecular orbital ( E LUMO) energy, E LUMO + E HOMO, and the average of net atomic charges on carbon atoms ( Q Cave) are higher, the corresponding alkylnaphthalene reaction rate constants would be higher. In contrast, higher values of the most positive net atomic charges on hydrogen atoms ( Q H + ) could lead to the decrease of log k values. H-atom abstraction may occur on the hydrogen atom with the highest atomic charge ( Q H). Radical addition reaction may also occur on the carbon (e.g. the carbon bonded to the alkyl group) with higher atomic charge ( Q C) values. Other descriptors such as molecular weight ( M w), standard heat of formation (Δ H f), total energy (TE), electronic energy (EE), core–core repulsion energy (CCR), average molecular polarizability ( a) and dipole moment ( μ) were also important descriptors in the QSPR/QSAR models.