This paper analyzes factors which determine the extent of density compensation on islands; i.e., is the summed population density of all species on an island equal to the summed mainland density? A graphical analysis allows quantitative comparisons of density compensation studies. Two hypotheses which are generally applicable predict the extent of density compensation on islands: (1) Niche theory predicts that summed population densities should be low if island species number is low. (2) The "habitat appropriateness" hypothesis predicts that summed population densities should be low if island populations occupy unfamiliar habitat. Both hypotheses successfully explain variability in the extent of density compensation on islands. Relative to the mainland, summed population densities on islands are high when islands support a large number of species and those species occupy familiar habitats. Summed population densities on islands are low when islands support few species and those species occupy novel habitats.