The kinetic energy release distributions (KERDs) associated with the hydrogen loss from the benzene cation and the deuterium loss from the perdeuteriobenzene cation have been remeasured on the metastable time scale and analyzed by the maximum entropy method. The experimental kinetic energy releases are larger than expected statistically, in contradistinction to what has been observed for the C-X fragmentations of the halogenobenzene cations. H(D) loss from C(6)H(6)(+) (C(6)D(6)(+)) occurs via a conical intersection connecting the (2)A(2) and (2)A(1) electronic states. Two models are proposed to account for the experimental data: (i) a modified orbiting transition state theory (OTST) approach incorporating electronic nonadiabaticity; (ii) an electronically nonadiabatic version of the statistical adiabatic channel model (SACM) of Quack and Troe. The latter approach is found to be preferable. It leads to the conclusion that the larger the energy stored in the transitional modes, which partly convert to the relative interfragment motion, the shorter the value of the reaction coordinate at which the adiabatic channels cross, and the larger the probability of undergoing the (2)A(2) --> (2)A(1) transition required for hydrogen loss.