Abstract For the mammalian faunas of 24 landbridge islands in the Gulf of Maine (0.003-279 km 2 in size), area accounts for 86% of variance in species richness. The slope, z, of the species-area curve is 0.247. For the seven largest islands (10km 2), the non-equilibrium hypothesis of relaxation following saturation in the post-Pleistocene is suggested by (1) elevated slope of the species-area curve (0.353), (2) correlation of species richness with island age ( r = −0.81) and water depth to mainland ( r = −0.70), (3) highly non-random nested subsets of species ranked by island area, and (4) discontinuity with the extremely depauperate faunas of oceanic islands of the eastern North Atlantic. The alternative hypothesis of a dynamic equilibrium determined by recurrent immigration and extinction is supported by (1) documented turnover in 16 species, (2) correlation of species-area residuals with distance ( r = − 0.90), (3) distribution dependent upon vagility with reduction or absence of hibernators and other poor dispersers, (4) low levels of endemism, and (5) congruence of community structure with that of mainland fauna for both trophic level and body size. I conclude that while some insular populations may be relictual, the faunal composition of most of these islands is dependent on recurrent colonization, much of which takes place over ice bridges. However, true equilibrium is perturbed by climatic shifts, range expansions, and human disturbance.