The so-called CO/Pt(111) puzzle, the experimentally demonstrated preference of CO to adsorb on the top site on the Pt(111) surface rather than the 3-fold hollow sites predicted by standard density functional theory (DFT) methods, was alleged to be solved by properly leveling the CO frontier molecular orbitals. However, the subtle energy difference between the top and hollow sites is of the same order of magnitude as the possible contribution of dispersive forces to this interaction. Herein, the role of dispersion in this system has been investigated by considering the PBE, PBEsol, RevPBE, RPBE, and SOGGA11 generalized gradient approximation (GGA) based exchange-correlation functionals; nonseparable functionals such as N12; and the TPSS and M06-L meta-GGA-type functionals, together with D2, D3, D3BJ, and MBD dispersion corrections. The results reinforce the recommendation of using M06-L for a correct description of CO adsorption-site preference even if including dispersion leads to a change of site and a noticeable overestimation of the adsorption energy indicating the presence of error-compensation effects. The present results also highlight the fact that dispersion contributes in bridging the preference gap between the top and hollow sites when other functionals are used. Dispersive forces play a role in site preference for CO on Pt(111), and it is likely that a similar situation is encountered for other late transition metals. Therefore, dispersion must be considered to reach a completely unbiased description of CO adsorption on metals. Nevertheless, including dispersion leads to adsorption energy values that overestimate the experimental value, indicating the limitations of the existing, widely used density functionals.