The contraction-expansion model (CEM) describes the dynamics of species that survived in refugia during the last glacial maximum (LGM) and expanded their range when environmental conditions slowly improved from the Late Glacial through to the Holocene. The CEM has been proposed to describe the dynamics of reef species in response to sea-level fluctuations from a range of disciplines, but genetic inferences rather suggest stable population sizes since the last glacial period. Here, we address this paradox by providing a new model of modern reef development, by assessing the effect of LGM bottlenecks using genetic simulations, and by using a survey of the literature on reef species to compile both estimates of times to expansion and applied rates of molecular evolution. Using previously published radiocarbon dates of core data, we propose a synthetic model for the dynamics of modern coral reefs in the Indo-Pacific region. This model describes both an initiation at 9.9 ka and subsequent development that confirms a strong influence of sea-level fluctuations on reef dynamics. Simulations based on mtDNA datasets showed that pre-LGM genetic signatures of expansion are lost. Recent literature shows that, although genetic expansions of tropical marine species are frequent (>95%), the onset of these expansions is old (median ~110 ka), which indicates that most populations have remained stable since before the LGM. These pre-LGM expansions are explained by the low mutation rates (1.66% changes/site/Myr) known to be inadequate to calibrate time at population level. Specific calibrations should help solve the paradox and generalise the CEM for reef species.