Schaefer's "J method" was employed to show that 2-cyclopropyl-1,3-dimethylbenzene (5) in solution prefers the perpendicular conformation in which the torsional angle Theta between the C(1)-H bond of the cyclopropyl group and the plane of the benzene ring is 90 degrees. This is opposed to the situation in cyclopropylbenzene (3) where the bisected conformer (Theta = 0 degrees) prevails. From the value of -0.85 +/- 0.01 Hz for (6)J(H-alpha,H-para) in 5 (for solutions in CS2 and in acetone) a barrier to rotation about the cyclopropyl-aryl bond of 6.4 kJ/mol can be derived if a predominantly two-fold potential and a vanishing (6)J(H,H) for Theta = 0 degrees are assumed. The introduction of the two ortho-methyl groups into 3 thus effectively interchanges the ground and transition state conformations of the internal rotation. This effect is well reproduced by ab initio (STO-3G and 6-31G*) and semiempirical (AM1) molecular orbital computations. The preference for the perpendicular conformation of an ortho-disubstituted cyclopropyl substituent was also demonstrated by a dynamic NMR study of 2-cyclopropyl-4-isopropyl-1,3,5-trimethy (10). Exchange-broadened H-1 spectra due to slow rotation of the cyclopropyl group were only obtained near the low-temperature limit of the spectrometer (-140 degrees C), and the barrier to rotation is estimated to lie near 28 kJ/mol (Delta G double dagger). MM3 molecular mechanics computations suggest that the rather large increase of the rotational barrier in 10 relative to 5 is caused by the combined buttressing effect of the isopropyl and the 5-methyl groups. The present findings explain why an earlier attempt (in 1970) to determine the rotational barrier in 2-cyclopropyl-1,3,5-trimethylbenzene by dynamic NMR was bound to fail.