Density-dependent mortality can regulate local populations – effectively minimizing the likelihood of local extinctions and unchecked population growth. It is considered particularly important for many marine reef organisms with demographically open populations that lack potential regulatory mechanisms tied to local reproduction. While density-dependent mortality has been documented frequently for reef fishes, few studies have explored how the strength of density-dependence varies with density, or how density-dependence may be modified by numerical effects (i.e., number-dependent mortality). Both issues can have profound effects on spatial patterns of abundance and the regulation of local populations. I address these issues through empirical studies in Moorea, French Polynesia, of the six bar wrasse (Thalassoma hardwicke), a reef fish that settles to isolated patch reefs. Per capita mortality rates of newly settled wrasse increased as a function of density and were well approximated by the Beverton-Holt function for both naturally formed and experimentally generated juvenile cohorts. Average instantaneous mortality rates were a decelerating function of initial densities, indicating the per capita strength of density-dependence decreased with density. Results of a factorial manipulation of density and group size indicate that per capita mortality rates were simultaneously density- and number-dependent; fish at higher densities and/or in groups had higher probabilities of disappearing from patch reefs compared with fish that were solitary and/or at lower densities. Mortality rates were ~30% higher for fish at densities of 0.5 fish/m2 than at 0.25 fish/m2. Similarly, mortality rates increased by ~45% when group size was increased from 1 to 2 individuals per patch, even when density was kept constant. These observations suggest that the number of interacting individuals, independent of patch size (i.e., density-independent effects) can contribute to regulation of local populations. Overall, this work highlights a greater need to consider numerical effects in addition to density effects when exploring sources of population regulation.