Coupled cluster (CC) and density-functional theory (DFT) calculations of optical rotation, [alpha](lambda), have been carried out for the difficult case of (S)-methyloxirane for comparison to recently published gas-phase cavity ringdown polarimetry data. Both theoretical methods are exquisitely sensitive to the choice of one-electron basis set, and diffuse functions have a particularly large impact on the computed values of [alpha](lambda). Furthermore, both methods show a surprising sensitivity to the choice of optimized geometry, with [alpha](355) values varying by as much as 15 deg dm(-1) (g/mL)(-1) among molecular structures that differ only negligibly. Although at first glance the DFT/B3LYP values of [alpha](355) appear to be superior to those from CC theory, the success of DFT in this case appears to stem from a significant underestimation of the lowest (Rydberg) excitation energy in methyloxirane, resulting in a shift of the first-order pole in [alpha](lambda) (the Cotton effect) towards the experimentally chosen incident radiation lines. This leads to a fortuitous positive shift in the value of [alpha](355) towards the experimental result. The coupled cluster singles and doubles model, on the other hand, correctly predicts the position of the absorption pole (to within 0.05 eV of the experimental result), but fails to describe correctly the shape/curvature of the ORD region lambda=355, resulting in an incorrect prediction of both the magnitude and the sign of the optical rotation. (C) 2004 American Institute of Physics.