A significant challenge in both measuring and predicting species extinction rates at global and local scales is the possibility of extinction debt, time-delayed extinctions that occur gradually following an initial impact. Here we examine how relative abundance distributions and spatial aggregation combine to influence the likely magnitude of future extinction debt following habitat loss or climate-driven range contraction. Our analysis is based on several fundamental premises regarding abundance distributions, most importantly that species abundances immediately following habitat loss are a sample from an initial relative abundance distribution and that the long-term, steady-state form of the species abundance distribution is a property of the biology of a community and not of area. Under these two hypotheses, the results show that communities following canonical lognormal and broken-stick abundance distributions are prone to exhibit extinction debt, especially when species exhibit low spatial aggregation. Conversely, communities following a logseries distribution with a constant Fisher's α parameter never demonstrate extinction debt and often show an "immigration credit," in which species richness rises in the long term following an initial decrease. An illustration of these findings in 25 biodiversity hotspots suggests a negligible immediate extinction rate for bird communities and eventual extinction debts of 30-50% of initial species richness, whereas plant communities are predicted to immediately lose 5-15% of species without subsequent extinction debt. These results shed light on the basic determinants of extinction debt and provide initial indications of the magnitude of likely debts in landscapes where few empirical data are available.