A combination of experimental and theoretical methods were employed to investigate the synthesis of methanol via CO₂ hydrogenation (CO₂ + 3H₂ → CH₃OH + H₂O) on Cu(111) and Cu nanoparticle surfaces. High pressure reactivity studies show that Cu nanoparticles supported on a ZnO(000) single crystal exhibit a higher catalytic activity than the Cu(111) planar surface. Complementary density functional theory (DFT) calculations of methanol synthesis were also performed for a Cu(111) surface and unsupported Cu₂₉ nanoparticles, and the results support a higher activity for Cu nanoparticles. The DFT calculations show that methanol synthesis on Cu surfaces proceeds through a formate intermediate and the overall reaction rate is limited by both formate and dioxomethylene hydrogenation. Moreover, the superior activity of the nanoparticle is associated with its fluxionality and the presence of low-coordinated Cu sites, which stabilize the key intermediates, e.g. formate and dioxomethylene, and lower the barrier for the rate-limiting hydrogenation process. The reverse water-gas-shift (RWGS) reaction (CO₂ + H₂ → CO + H₂O) was experimentally observed to compete with methanol synthesis and was also considered in our DFT calculations. In agreement with experiment, the rate of the RWGS reaction on Cu nanoparticles is estimated to be ∼2 orders of magnitude faster than methanol synthesis at T = 573 K. The experiments and calculations also indicate that CO produced by the fast RWGS reaction does not undergo subsequent hydrogenation to methanol, but instead simply accumulates as a product. Methanol production from CO hydrogenation via the RWGS pathway is hindered by the first hydrogenation of CO to formyl, which is not stable and prefers to dissociate into CO and H atoms on Cu. Our calculated results suggest that the methanol yield over Cu-based catalysts could be improved by adding dopants or promoters which are able to stabilize formyl species or facilitate the hydrogenation of formate and dioxomethylene.