The relative importance of predators and resources (i.e., food) for the dynamics of migratory bird populations is poorly known. Resource availability may be more likely in resource poor environments, but given that nest failure in most systems is due mainly to predation, predator effects may predominate. We document a rapid decline of an isolated Eastern Kingbird (Tyrannus tyrannus) population breeding in the Great Basin Desert of eastern Oregon, USA, and evaluate whether it was driven by limited food resources (water availability ~ food), nest predation, or first-year or adult return rate (RR_J and RR_A, respectively) that reflect nonbreeding season events. Most nests failed (~68% of nests) due mainly to nest predation (>90% of failures); nestling starvation was rare. Bioyear precipitation (October-April), breeding season precipitation, and river flow all varied widely but none could account for annual variation in either nest success (NS) or fledging success of successful nest (FSSN). Neither RR_J nor RR_A varied with year, any measure of water resources, or reproductive success. Annual population growth rate (ln[N_t+1/N_t]) was independent of year, all measures of water availability, RR_J and RR_A, and NS of replacement nests, but was high following years of high NS of initial nests; FSSN was also higher in years of high NS. High rates of nest predation, due likely to the expansion of the local American Crow (Corvus brachyrhynchos) population, appeared to drive the population decline. Empirical data and population simulations indicated that replacement nests contributed little to λ, while simulations revealed that, given site-specific vital rates, the decline would have been even greater without immigration (λ = 0.812) than the observed (λ = 0.914). Long distance dispersal may thus be critical for persistence of kingbirds in the naturally highly fragmented habitat that it occupies in the Great Basin Desert.