As revealed by several experimental examples, radicals and ion radicals may, in contrast with closed-shell molecules, undergo exothermic homolytic cleavages (.A..B --> A: + .B) with substantial activation barriers. A two-state semiclassical model is proposed for explaining the existence of the barrier and estimating its magnitude. It is based on the intersection of the potential energy surfaces characterizing the dissociation of a bonding state, .A..B --> .A. + .B, on one hand, and the approach to bonding distance of a repulsive state, A: + .B --> AthereforeB, on the other. After inclusion of the bond cleavage and formation as Morse curves in the normal-mode analysis, a simple activation driving force relationship is obtained, the two main ingredients of the intrinsic barrier being the triplet excitation energy of the A moiety and the pi* --> sigma* excitation energy in .A-B. The model is then tested by quantum chemical calculations, first on a simplified system to evaluate the calculation techniques and then on a real system. A comparison of the model predictions with experiment is finally performed using the rate data recently gathered for the cleavage of 4-cyanophenyl alkyl ether anion radicals, which cover a respectable range of driving forces, showing satisfactory agreement between theoretical predictions and experimental data.