SUMMARYIn order to better understand community-level effects of fishing on temperate reefs at continental scales, changes in densities of common species in five Australian marine protected areas (MPAs) were estimated from prior to establishment to three years after enforcement of fishing prohibitions. A before-after-control-impact survey design was used, with 5–14 replicated sites distributed within both sanctuary and fishing zones associated with each MPA. On the basis of published meta-analyses, exploited species were generally expected to show increased densities. By contrast, only two of the 11 exploited fish species (the red morwongCheilodactylus fuscusand latrid trumpeterLatridopsis forsteri), and none of seven exploited invertebrate species, showed significant signs of population recovery within sanctuary zones. Four fish species increased in biomass between survey periods. When variation in abundance data was partitioned by PERMANOVA independently for the five MPAs, the ‘zone × year’ interaction component consistently contributed onlyc.4% of total variation, compared to site (c.35%), zone (c.8%), year (c.8%) and residual error (c.45%) components. Given that longer-term Australian studies show clear community-wide responses following MPA protection, the discrepancy between weak observed recovery anda prioriexpectations is probably due, at least in part, to the three-year period studied being insufficient to generate clear trends, to relatively low fishing pressure on some temperate Australian reefs, and to meta-analyses overestimating the likelihood of significant short-term population responses.