Two molecular systems for which previous studies had found qualitative differences in the results from calculations with the B3LYP and MP2 methods are investigated with a range of different electronic structure methods, including meta and double hybrid density functionals and DFT-D (DFT augmented with an empirical dispersion term). The performance of the different methods is assessed by comparison to estimated CCSD(T)/CBS (complete basis set) results. The first molecular system studied is the ethyne methyl isocyanide complex (CH3NC-C2H2), which exhibits π hydrogen bonds involving the C≡C and N≡C triple bonds. Earlier work on this system had shown that B3LYP predicts significantly longer hydrogen-bond distances than MP2. Here, we show that this is likely due to missing dispersion in the B3LYP calculations. On the basis of the CCSD(T) results, the ethyne methyl isocyanide interaction energy is estimated to be 12 ± 1 kJ/mol. B3LYP significantly underestimates the stability of the complex, whereas MP2 slightly overestimates. M05-2X, B3LYP-D, and (CP-corrected) mPW2-PLYP-D give results in close proximity to the CCSD(T) reference values. The second molecule investigated is thioanisole (C6H6SCH3), which can adopt two different conformations (thiomethyl group either planar or perpendicular with respect to the benzene ring). Potential energy curves for rotation around the C(sp(2))-S bond are computed and compared to the estimated CCSD(T)/CBS curve. CCSD(T) predicts the planar conformation to be the global minimum, with a plateau region near the perpendicular conformation (∼4 kJ/mol higher in energy than the planar conformation). The shape of the curve, and location of minima and barriers, is very dependent on the method and basis set employed. MP2, B3LYP, M05-2X, mPW2-PLYP, and mPW2-PLYP-D (employing basis sets of double- or triple-ζ quality) give results in reasonable agreement with the CCSD(T) results, whereas B3LYP-D and M06-L give vastly overestimated barriers at the perpendicular conformation.