Reactions of Fe+ and Ni+ with propane, propane-2-d(1), propane-2,2-d(2), propane-1,1,1-d(3), propane-1,1,1,3,3,3-d(6) and propane-d(8) are examined to gain insight into the mechanism and energetics for the H-2 and CH4 elimination channels. The questions of C-H and/or C-C bond activation and the relative contributions from primary and secondary C-H bond activation are addressed. Total cross section measurements indicate that ground-state Ni+(D-2) and Fe+(D-6) react with propane inefficiently, 13% and 7.5% of the Langevin collision cross section, respectively, with CH4 loss favored over H-2 loss by a factor of 4.0 for Ni+ and 2.8 for Fe+. For reactions with C3D8, the total cross sections decrease by factors of 3.8 for Ni+ and 4.4 for Fe+ relative to C3H8, with the dehydrogenation channel enhanced over demethanation for both Ni+ and Fe+. Kinetic energy release distributions (KERDs) from nascent metastable Ni(propane)(+) and Fe(propane)(+) complexes were measured for H-2 loss and CH4 loss. For H-2 loss, the distribution is bimodal. Studies using propane-2,2,d(2) and propane-1,1,1,3,3,3-d(6) indicate that both primary and secondary C-H insertions are involved as initial steps. Initial secondary C-H insertion is responsible for the high-energy component in the bimodal KERD, which is much broader than predicted from statistical theory, indicating that a tight transition state leads to the final products. The low-energy component for H-2 loss involves initial primary C-H insertion and appears to be statistical, suggesting little or no reverse activation barrier as the system separates to products. The kinetic energy distribution for demethanation is statistical and is very sensitive to the energy of the rate-limiting C-H insertion transition state. A lower limit for the energy of this transition state is obtained by modeling the experimental kinetic energy release distribution for demethanation using statistical phase space theory. The barrier reduces the contribution of high angular momentum states to the final products, thus reducing the high-energy portion of the product kinetic energy distribution. Modeling the cross section, the isitope effect, and the KERD for CH4 loss using statistical phase space theory indicates that the barrier for C-H bond insertion is located 0.10 +/- 0.03 eV below the Ni+/C3H8 asymptotic energy and 0.075 +/- 0.03 eV below the Fe+/C3H8 ground-state asymptotic energy. All data can be explained by initial C-H insertion, without the need to invoke initial C-C bond activation for ground-state Fe+ and Ni+ reacting with propane at low kinetic energy.