In the study we used a number of high level theoretical methods to calculate the O-H bond dissociation energies (BDEs) as well as a and remote para substituent effects on them. We found that only G3 and CBS-Q methods can be used to calculate the absolute O-H BDEs. Other methods including B3LYP, MP2, and CCSD(T), either open-shelled or close-shelled, significantly underestimate the O-H BDEs. To be even worse, except for G3 and CBS-Q. the other theoretical methods cannot accurately predict the a substituent effects on O-H BDEs either. Methods including UMP2, ROMP2, and UCCSD(T) may even provide qualitatively erratic alpha substitutent effects. Using the G3 and CBS-Q results, we found that the alpha substituent effect on the O-H BDEs is usually much larger than that on the C-H or N-H BDEs. Both the pi donors and acceptors reduce the O-H BDEs because of the conjugation and hyperconjugation between the substituent and the radical center. Polyfluorinated alkyl groups increase the O-H BDEs because of the inductive electron-withdrawing effect. In comparison, an electron-withdrawing para substituent increases the O-H BDE of phenol, whereas an electron-donating group reduces it. The calculated rho(+) value for the O-H BDEs of phenols is about 4-5 kcal/mol. Compared to it, the experimentally determined rho(+) value is significantly larger because of the solvent effect. Furthermore, the rho(+) values for the O-Y BDEs of 4-X-C6H4-O-Y decrease in the order O-CH3 > O-H > O-OCH3 > O-OH > O-NO. (C) 2003 Elsevier B.V. All rights reserved.