In 1998, seawater temperature anomalies led to unprecedented levels of coral bleaching on reefs worldwide. We studied the direct effects of this thermal event on benthic communities and its indirect effects on their associated coral reef fish communities at a group of remote reefs off NW Australia. Long-term monitoring of benthic and fish assemblages on these reefs allowed us to compare the responses of these communities to coral bleaching using a data series that included 4 years before, and 6 years following, this bleaching event. While bleaching mortality was evident to >30 m depth, it was patchy among the shallower survey sites with decreases in live coral cover ranging from 30% to 90% across seven surveyed locations Within 2 years of the bleaching, hard coral recovery had begun at all sites and by 2003 reef-wide coral cover had increased to ∼39% of its preimpact levels. We exploited this pattern of differential survival of corals among sites, the associated changes in these benthic communities, and their patterns of recovery, to better understand links between benthic community dynamics and their associated fish communities. Temporal changes in the resident fish communities strongly reflected the differential shifts in the benthic communities, but were lagged by 12–18 months. Five years after the bleaching event, the fish communities on five of the seven surveyed locations showed evidence of recovery, however, none had regained their preimpact structures. Analyses of these communities by taxonomic family revealed a range of responses to the disturbance reflective of their life-histories and trophic and habitat affiliations. The slow but recognizable recovery of this isolated reef system has parallels with other relatively isolated systems that displayed resilience to the 1998 bleaching event, e.g. the Chagos archipelago, but it also contrasts sharply with low levels of resilience documented in other isolated reef systems subject to the same disturbance, e.g. the Seychelles. In this context, our results highlight the significant knowledge gaps remaining in understanding the resilience of these ecosystems to disturbance.