We compare CO’s response to electric fields at edge and terrace sites on the stepped Pt(335) surface. The comparison is made at zero frequency and at the frequency of the C-O stretch vibration. Atop-bonded CO is observed with reflection-absorption ir spectroscopy (RAIRS), electroreflectance vibrational spectroscopy (EVS), and high-resolution electron energy loss spectroscopy (HREELS). Coadsorbed H or O is used to control the CO adsorption site. With both RAIRS and HREELS the measured vibrational cross-section of atop CO at the step edge is 2.0+/-0.2 times greater than for CO on the terrace. The vibrational Stark effect-the change of CO’s vibrational frequency with externally applied electrostatic field-is also a factor 2.0+/-0.2 larger for atop CO at the step edge than it is for atop CO on the terrace. Because the vibrational cross section varies as the square of the field while the Stark effect is linear, a model in which CO responds to the screened local field at a single point cannot simultaneously explain the observed site dependence of both the vibrational Stark effect and the vibrational cross section. The most plausible explanation is that CO’s response to electric fields is nonlocal. A simple model is presented that can account for our data.