Abstract Early studies stressed the high accretionary potential of Caribbean coral reefs (10-15 m/kyr). Cores from St. Croix (US Virgin Islands) suggested that the earliest reefs along the shelf edge were dominated by rapidly growing Acropora palmata. As the bank flooded, soil mobilized by wave action moved off the bank, killed these reefs, and prevented subsequent reef development for at least 2000 years. Because the shelf east of St. Croix is similar to many Caribbean sites, it was proposed that this model was widely applicable throughout the region. More recent data from cores and outcrops throughout the Caribbean show that reefs on St. Croix and elsewhere flourished throughout the hiatus originally proposed to have occurred between 10,000 and at least 8000 years ago. Furthermore, data compiled from all known Caribbean cores suggest that reefs built at rates averaging only 3.47 m/kyr and that reef-accretion rates above 7 m/kyr were exceedingly rare. The pattern of reef building and abandonment over the past 20,000 years is consistent with these findings. Reefs easily kept up with sea level when it was rising at rates below 3-5 m/kyr, and back-stepping consistently occurred throughout the intervening interval when sea level rose at rates up to 10 m/kyr. While sudden shifts in sea level have been well documented in the geologic record and degraded water quality will undoubtedly compromise reef building, triggers for back-stepping need not be confined to these scenarios. The mechanisms for back-stepping are complicated and still poorly understood. In virtually every well-documented case for the Holocene, neither community structure nor water depth over the abandoned reefs and their landward replacements were significantly different. Each new reef formed as sea level rose over an antecedent feature that favored reef formation, and its success was largely determined by the rate of sea-level rise at the time accretion began. Reefs that formed when sea level was rising faster than 3-4 m/kyr were eventually abandoned. The scale of these reefs and the magnitude of back-stepping are similar to many ancient examples. Assumptions that these reflect sudden shifts in sea level or severe environmental conditions should be reexamined in light of the realization that back-stepping can also be explained by physical and biological processes that are observable within the lifetime of a single reef scientist.