Numbers and biomass of piscivorous fish and their predation on other fish may often be high in undisturbed coral reef communities. The effects of such predation have sometimes been studied by removal of piscivores (either experimentally or by fishermen). Such perturbations have usually involved removal of large, highly vulnerable, mobile piscivores that are often actively sought in fisheries. The effects of fishing on smaller, demersal, semi-resident piscivores have been little studied. We studied such effects on the fish communities of patch reefs at Midway atoll by experimentally removing major resident, demersal, piscivorous fishes. First, four control reefs and four experimental reefs were selected, their dimensions and habitats mapped, and their visible fish communities censused repeatedly over 1 year. Census of all control and experimental reefs was continued for the following 39 months, during which known piscivores were collected repeatedly by hand spearing. Records were kept of catch and effort to calculate CPUE as an index of predator density. Spearfishing on the experimental reefs removed 2504 piscivorous fish from 12 families and 43 taxa (mostly species). The species richness of the catch did not show an overall change over the duration of the experiment. Spearman rank correlation analysis showed some unexpected positive correlations for density in numbers and biomass of major fished piscivorous groups (especially lizardfish) over the experiment. Only two relatively minor fished piscivorous taxa declined in abundance over the experiment, while the overall abundance of piscivores increased. Visual censuses of fish on the experimental reefs also failed to show reduction of total piscivores over the full experimental period. No significant trend in the abundance of lizardfish censused over the full period was apparent on any of the control reefs. The high resilience of piscivores on these experimental reefs to relatively intense fishing pressure could result from their protracted recruitment seasons, high immigration rates, cryptic habits, or naturally high abundances. A major factor was the high immigration rates of lizardfish, replacing lizardfish and other less mobile piscivores removed from the reefs by spearing. On the fished reefs, the removed lizardfish population replaced itself >20 times during the experiment; other piscivorous taxa replaced themselves only 5 times.