Rotaxanes that can be switched between co-conformations by some external stimulus are of interest because the switching mechanism might be used to create molecular devices capable of producing useful work. Probably the most common approach to create a switchable rotaxane is to start with a rotaxane where the ring interacts more strongly with one of two possible binding sites along the shaft, then apply an external stimulus that weakens the binding interaction between the ring and the shaft at this site, thereby changing the binding site preference. We have investigated this binding site preference with electronic structure calculations. To gain insight into the origins of the inter-component binding, electronic structure calculations were carried out at several levels of theory and analytic approximations were applied to estimate the electrostatic and dispersion contributions. Dispersion has been thought to make an important contribution to the inter-component interaction in the presence of p-p stacking interactions between the components, but the role of dispersion interaction has been a controversial issue because many computational methods neglect this interaction. For example, AM1 semiempirical calculations neglect dispersion, but typically predict the correct co-conformational preferences. This suggests that inclusion of the dispersion interaction is required for correct quantitative, but not qualitative, description of the inter-component binding, a result that is supported by the analytic partitioning of the binding interactions.