Phenoxy radicals are expected to play an important role in the thermal decomposition of lignin. At high enough temperatures, the dissociation to cyclopentadienyl radicals plus carbon monoxide dominates over bimolecular channels. While earlier experimental and theoretical studies agree on the mechanism, the experimental data suggest a substantially lower overall barrier than that found in theoretical studies. To address this discrepancy, we performed electronic structure calculations at the CBS-QB3 level of theory and at a modified version of it to construct the relevant portion of the C6H5O potential energy surface (PES). The new results are in good agreement with previous studies. We then used the molecular information to calculate thermodynamic properties of the reactants and activated complexes, the high-pressure rate constants, and the pressure dependence. Three different methods were employed for the last step: the quantum version of the Rice–Ramsperger–Kassel/modified strong collision theory approach by Dean (J Phys Chem 1985, 89, 4600–4608), the stochastic treatment by Barker (Int J Chem Kinet 2001, 33, 232–245) using the MultiWell package, and the master equation program Unimol by Gilbert and coworkers. Theory of Unimolecular and Recombination Reactions. Oxford, Blackwell Scientific Publications. All methods predict the experimentally determined phenoxy decomposition rate constant to be in the falloff region. This explains the almost 10 kcal mol−1 difference between reported activation energies and calculated barriers. Both the Lin and Lin (J Phys Chem 1986, 90, 425–431) and Frank et al. (Proc Combust Inst 1994, 25, 833–840) data can be reproduced within the assumed uncertainty limits without any adjustments of the PES. We also report on falloff behavior in the CO elimination from methyl-, hydroxy-, and methoxy-substituted phenoxy radicals in the para position and compare it to that of the unsubstituted phenoxy radical. © 2011 Wiley Periodicals, Inc. Int J Chem Kinet 44: 75–89, 2012