Temperature-programmed desorption (TPD) was used to investigate the desorption of butane, hexane, and octane from Al2O3(0001) in ultrahigh vacuum. At low coverages, TPD traces for butane and hexane displayed one peak which was attributed to monolayer desorption. A second, multilayer peak, was observed at a lower temperature as the coverage was increased. However, the multilayer peak appeared at coverages well below the saturation coverage of the monolayer peak implying that the multilayer was forming before the monolayer was completely full. A simple statistical adlayer sticking model was used fo simulate the relative number of molecules in the monolayer and multilayer as a function of total coverage giving good agreement with the TPD data. In addition, the variation of ramp rates method was used to measure the desorption kinetics at coverages well below one monolayer for each of the alkanes. All three alkanes displayed first-order desorption kinetics with activation barriers of butane E(d) = 8.4 +/- 1.2 kcal/mol; hexane E(d) = 10.4 +/- 0.8 kcal/mol; octane E(d) = 14.6 +/- 0.8 kcal/mol. The first-order preexponentials were butane v(1) = 4 x 10(10+/-2) s(-1); hexane v(1) = 5.4 x 10(9+/-1.5) 5 s(-1); octane v(1) = 1.6 x 10(12+/-2) s(-1). The comparison of these desorption barriers to the bulk heats of sublimation along with the separation between monolayer and multilayer peaks in the TPD as a function of chain length suggest that the relative magnitude of molecule-surface interactions compared to molecule-molecule interactions decreases with increased chain length.