Theoretical estimates of long-term minimum viable population (MVP) sizes for mammals indicate MVPs of tens of thousands, even a million. However, data to test the theoretical estimates are effectively non-existent. I here use information on distributions of primates on islands of southeast Asia to provide empirical estimates of the size of multi-millennial MVPs following the post-Pleistocene insularization in the Sunda region. Small Sunda region islands have fewer taxa than do large ones. Assuming extinction on the small islands, the MVP size is somewhere between the population size on the smallest island on which the taxon is extant and that on the largest island on which the taxon no longer exists (population size calculated as density of taxon by area of land available). Results for eight to ten genera (depending on the taxonomy) on 35 islands indicate MVPs of perhaps a few score for Nycticebus (loris) and Macaca (macaque) to several thousands, or scores of thousands, for Pongo (orang-utan) and Symphalangus (siamang). Large-bodied taxa have significantly larger MVPs than do small-bodied taxa. Only four protected areas in southeast Asia are large enough to conserve the smaller MVPs of the siamang and orang-utan.