Abstract Understanding the fitness consequences of inbreeding (inbreeding depression) is of importance to evolutionary and conservation biology. There is ample evidence for inbreeding depression in captivity, and data from wild populations are accumulating. However, we still lack a good quantitative understanding of inbreeding depression and what influences its magnitude in natural populations. Specifically, the relationship between the magnitude of inbreeding depression and environmental severity is unclear. We quantified inbreeding depression in survival and reproduction in populations of cactus finches (Geospiza scandens) and medium ground finches (Geospiza fortis) living on Isla Daphne Major in the Galapagos Archipelago. Our analyses showed that inbreeding strongly reduced the recruitment probability (probability of breeding given that an adult is alive) in both species. Additionally, in G. scandens, first-year survival of an offspring withf= 0.25 was reduced by 21% and adults withf= 0.25 experienced a 45% reduction in their annual probability of survival. The magnitude of inbreeding depression in both adult and juvenile survival of this species was strongly modified by two environmental conditions, food availability and number of competitors. In juveniles, inbreeding depression was only present in years with low food availability, and in adults inbreeding depression was five times more severe in years with low food availability and large population sizes. The combination of relatively severe inbreeding depression in survival and the reduced recruitment probability led to the fact that very few inbred G. scandens ever succeeded in breeding. Other than recruitment probability, no other trait showed evidence of inbreeding depression in G. fortis, probably for two reasons: a relatively high rate of extrapair paternity (20%), which may lead to an underestimate of the apparent inbreeding depression, and low sample sizes of highly inbred G. fortis, which leads to low statistical power. Using data from juvenile survival, we estimated the number of lethal equivalents carried by G. scandens, G. fortis, and another congener, G. magnirostris. These results suggest that substantial inbreeding depression can exist in insular populations of birds, and that the magnitude of the inbreeding depression is a function of environmental conditions.