Calculations of the desorption of hydrogen from Pt(110)-(1×2), a surface used to model nanoparticle edge sites, show the activation energy varying strongly with hydrogen coverage, from 0.8 to 0.3 eV. The predicted temperature programed desorption spectra agree well with experiments, but the formation of the hydrogen molecules occurs only at two types of sites on the surface even though three peaks are observed. The lowest and highest temperature peaks result from desorption from the same strong binding sites at the ridge, while desorption from the weakest binding trough sites is insignificant.