The recently proposed scheme of Grimme (BE scheme) (J. Ant. Chem. Sec. 1996, 118, 1529) to calculate intrinsic bond energies (BE's) of hydrocarbons, which define seminal equilibrium quantities of chemical structures, is evaluated critically. CH and regular CC bonds are treated well; the corresponding BE's are reliable and self-consistent. In contrast, the performance of the method is markedly reduced for bonds of unusual length, if the bond length is not determined by bond bending or by conjugation. Differences between BE's for CH bonds, which Lie within the remarkably narrow range from ca. 103 to 110 kcal mol(-1), and CH bond dissociation energies (BDE's, ca. 86-132 kcal mol(-1), linear correlation, R-c = 0.9291) give a measure of radical (de)stabilization. BE's of sp(x)-sp(y) CC single bonds correlate linearly with the respective BDE's (R, = 0.9987) and can be used for a reliable prediction of BDE's at almost no computational cost. Individual intrinsic bond energies are used to establish CC and CH bond length-bond energy-bond order correlations. In extension of Grimme's original report, the performance of the model is tested thoroughly for anions, cations, and radicals of hydrocarbons and it is shown that these species are treated less satisfactorily. Attempts to treat non-hydrocarbon compounds by the same procedure are also less successful with the exception of saturated silicon hydrides. Results of this work show that the relationships between bond length-bond order-bond energy as described by established models of the chemical bond can be related to the properties of the electron density at bond critical points. Despite the much greater angle distortion, cyclopropane has a strain energy only slightly larger than cyclobutane. This problem of the nearly equivalent strain energies is readdressed, leading to new estimates for the stabilization of cyclopropane due to CH bond strengthening (11.7 kcal mol(-1)) and to sigma -aromaticity (11.3 kcal mol(-1)).