A literature survey on the kinetics of hydride abstractions from CH-groups by carbocations reveals a general phenomenon: Variation of the hydride acceptor affects the rates of hydride transfer to a considerably greater extent than an equal change of the thermodynamic driving force caused by variation of the hydride donor. The origin of this relationship was investigated by quantum chemical calculations on various levels of ab initio and DFT theory for the transfer of an allylic hydrogen from 1-mono- and 1,1-disubstituted propenes (XYC=CH-CH(3)) to the 3-position of 1-mono- and 1,1-disubstituted allyl cations (XYC=CH-CH(2)(+)). The discussion is based on the results of the MP2/6-31+G(d,p)//RHF/6-31+G(d,p) calculations. Electron-releasing substituents X and Y in the hydride donors increase the exothermicity of the reaction, while electron-releasing substituents in the hydride acceptors decrease exothermicity. In line with Hammond's postulate, increasing exothermicity shifts the transition states on the reaction coordinate toward reactants, as revealed by the geometry parameters and the charge distribution in the activated complexes. Independent of the location of the transition state on the reaction coordinate, a value of 0.72 is found for Hammond-Leffler's alpha = deltaDeltaG/deltaDelta(r)G degrees when the hydride acceptor is varied, while alpha = 0.28 when the hydride donor is varied. The value of alpha thus cannot be related with the position of the transition state. Investigation of the degenerate reactions XYC=CH-CH(3) + XYC=CH-CH(2)(+) indicates that the migrating hydrogen carries a partial positive charge in the transition state and that the intrinsic barriers increase with increasing electron-releasing abilities of X and Y. Substituent variation in the donor thus influences reaction enthalpy and intrinsic barriers in the opposite sense, while substituent variation in the acceptor affects both terms in the same sense, in accord with the experimental findings. Marcus theory is employed to treat these effects quantitatively.