The abundant-centre hypothesis predicts that species' abundances peak at the centre of their geographical ranges and decline gradually towards their range limits. We tested predictions of this hypothesis for three rocky-shore, intertidal invertebrates with planktonic larvae (the whelk, Morula marginalba, the snail, Afrolittorina pyramidalis, and the barnacle, Tesseropora rosea) by quantifying their patterns of abundance and size, and inferring pulses of recruitment from size-frequency distributions, at multiple spatial scales spanning a 600-km region in south-eastern Australia and encompassing roughly the southern third of their geographical ranges. At the regional scale, abundances for all species were, as predicted, dramatically lower at their range limits. This decline was not gradual, however, because there were large variations in abundance at smaller spatial scales, and abrupt declines at the south-eastern corner of Australia. Size did not change towards the range limit for any species, but size-frequency distributions suggested a decline in the frequency of recruitment events at the range limit for T. rosea. We conclude that the abundant-centre hypothesis is not an appropriate model for abundance distributions of benthic marine invertebrates with planktonic larvae, because of the vagaries of dispersal and recruitment interacting with complex current patterns along non-uniform coastlines.