Habitat area and fragmentation are recognised as important factors that influence population size, community structure and extinction risk. Abundance and species richness universally increase with habitat area. However, the effects of different aspects of habitat fragmentation, including variation in patch size, number and isolation are often not distinguished from each other or the overall effects of habitat amount. Here we experimentally tested predictions concerning the effects of isolation on abundance, species richness and community structure of coral reef fishes colonising patch reefs by constructing clusters of patches of the same number and size, but manipulating reef spacing. Hexagonal clusters of 7 experimental patch reefs (6 edge and 1 central) with 3 levels of isolation (1 m, 5 m, and 15 m spacing) were established at Kavieng, Papua New Guinea and colonisation was recorded after 6 weeks in 2014. We also deployed video cameras to test whether isolation affected the activity of transient predatory fishes. As predicted, isolation had a positive effect on both mean abundance and species richness at both the cluster and patch scale. The cumulative abundance and species richness exhibited linear increases in relation to habitat area within clusters (from 1 to 7 patch reefs), but the slope increased with the degree of isolation. There was some evidence that transient predators remained longer and were more successful when patches were close together, which may explain the lower abundance and richness of juvenile fish assemblages on more aggregated patch reefs. This study demonstrates that while habitat amount is fundamentally important, isolation has significant effects that will need to be distinguished from other aspects of fragmentation when examining the processes structuring reef fish communities.