Valence shell electron pair repulsion theory (VSEPR) as explained in most textbooks predicts that substituents bonded to a central atom in AX[subscript n]E[subscript z][superscript c] species (A = main-group central atom, X = substituent, E = lone pair on central atom, c = charge) will change their X-A-X angles to bend away from the lone pairs. Exceptions have appeared in the literature, commonly arising from steric repulsions between very large substituents and less commonly from electronic factors such as multiple bonding and bond polarization. We have conducted extensive computational studies of hypercoordinate main-group molecules and ions AX[subscript n]E[subscript z][superscript c] and AO[subscript m]X[subscript n]E[subscript z][superscript c], where X = halide, and found that VSEPR-based predictions of such bending for those species containing heavier halides are likely incorrect. Indeed, despite the fact that cases where X = F usually conform to the prediction, we find that IOF[subscript 4][overbar]/XeOF[subscript 4] and IO[subscript 2]F[subscript 2][overbar]/XeO[subscript 2]F[subscript 2] should not. Calculations of the electron localization function indicate that the root cause of the difference is the migration of lone pairs closer to the central atom. We recommend that presentation of VSEPR in general chemistry and inorganic chemistry textbooks be revisited and provide suggested language incorporating this phenomenon.