Electrochemical experiments find that an adsorbate's vibrational frequency nu varies as a function of electrode potential phi. This has been attributed both to the effect of phi on the molecule's bonding and to the vibrational Stark effect-the molecule's interaction with local electrostatic field F. Both theories are shown to describe the same effect. In application, however, they involve different approximations. For CO on Pt, (d nu/d phi) with aqueous electrolyte is the same as (Delta nu/Delta Phi) in vacuum, where Delta Phi, is the coadsorbate-induced change of the metal's work function. This is explained in terms of CO's bonding. With a nonaqueous electrolyte, the effect of cation size on (d nu/d phi) is explained by the vibrational Stark effect. At low CO coverage in vacuum, a semiclassical theory accurately predicts (d nu/d phi). Ab initio calculations of (d nu/d phi) are also discussed. For CO on two surfaces, Pt(lll) and Pt(335), both (d nu/d phi) with aqueous electrolyte and (d nu/dF) in vacuum have been measured. If local F is assumed to be screened the same by the metal's electrons in electrolyte and vacuum, a comparison gives (dF/d phi) = 0.59 +/- 0.07 and similar to 0.85 Angstrom for Pt(111) and Pt(335), respectively. This is significantly different from the (dF/phi) similar to 0.28 Angstrom(-1) predicted by simple models of the double layer.