Human impacts on the world's oceans have been substantial, leading to concerns about the extinction of marine taxa. We have compiled 133 local, regional and global extinctions of marine populations. There is typically a 53-year lag between the last sighting of an organism and the reported date of the extinction at whatever scale this has occurred. Most disappearances (80%) were detected using indirect historical comparative methods, which suggests that marine extinctions may have been underestimated because of low-detection power. Exploitation caused most marine losses at various scales (55%), followed closely by habitat loss (37%), while the remainder were linked to invasive species, climate change, pollution and disease. Several perceptions concerning the vulnerability of marine organisms appear to be too general and insufficiently conservative. Marine species cannot be considered less vulnerable on the basis of biological attributes such as high fecundity or large-scale dispersal characteristics. For commercially exploited species, it is often argued that economic extinction of exploited populations will occur before biological extinction, but this is not the case for non-target species caught in multispecies fisheries or species with high commercial value, especially if this value increases as species become rare. The perceived high potential for recovery, high variability and low extinction vulnerability of fish populations have been invoked to avoid listing commercial species of fishes under international threat criteria. However, we need to learn more about recovery, which may be hampered by negative population growth at small population sizes (Allee effect or depensation) or ecosystem shifts, as well as about spatial dynamics and connectivity of subpopulations before we can truly understand the nature of responses to severe depletions. The evidence suggests that fish populations do not fluctuate more than those of mammals, birds and butterflies, and that fishes may exhibit vulnerability similar to mammals, birds and butterflies. There is an urgent need for improved methods of detecting marine extinctions at various spatial scales, and for predicting the vulnerability of species.