Temporal variation in survival, fecundity, and dispersal rates is associated with density-dependent and density-independent processes. Stable natural populations are expected to be regulated by density-dependent factors. However, detecting this by investigating natural variation in density is difficult because density-dependent and independent factors affecting population dynamics may covary. Therefore, experiments are needed to assess the density dependence of demographic rates. In this study, we investigate the effect of density on demographic rates of the Seychelles Warbler (Acrocephalus sechellensis). This species, endemic to a few islands in the Indian Ocean, went through a severe population bottleneck in the middle of the last century, with only ∼30 individuals left on one small island, but has since recovered. Our monitoring shows that since reaching the island's carrying capacity, population density has remained stable. However, we detected neither density-dependent reproduction nor survival on the basis of natural density variation during this stable period. For conservation reasons, new populations have been established by transferring birds to nearby suitable islands. Using the change of numbers during the process of saturation as a natural experiment, we investigated whether we can detect regulation of numbers via density-dependent survival and reproduction within these new populations. We found that populations were mainly regulated by density-dependent reproduction, and not survival. Variation in density between islands can be explained by food abundance, measured as insect density. Islands with the highest insect densities also had the highest bird densities and the largest breeding groups. Consequently, we suggest that the density-dependent effect on reproduction is caused by competition for food.