Nest box provisioning is a common management tool intended to increase population size or stability of threatened birds, but its effectiveness is rarely assessed. The provisioning of nest boxes may lead to unexpected results if nest type imprinting prevents naïve adult birds from immigrating into the nest box population, or limiting the ability of juveniles reared in nest boxes to emigrate to areas with only natural nesting substrates. We analyzed the population trends from 2008 to 2010 of southeastern American kestrels Falco sparverius paulus associated with a network of nest boxes in north-central Florida, USA, with Bayesian integrated population models (IPMs) that simultaneously considered mark–recapture data sets, fledgling production, and population surveys. We evaluated the demography of the nest box population by comparing population growth rates, apparent survival probabilities, and recapture probabilities between an IPM that explicitly modeled immigration and one that did not. Overall population growth rates suggested that the population was stable, and that immigration was apparently important in maintaining this stability, with approximately 0.3 and 0.5 female immigrants per resident female kestrel each year. Explicitly modeling immigration resulted in lower estimates of juvenile kestrel apparent survival probability, suggesting that a large proportion of locally produced juveniles emigrated rather than recruited locally. We concluded that neither preference for natural cavities nor imprinting on artificial nest boxes appeared to prevent immigration from maintaining the stability of the local population. Natal habitat preference imprinting on nest sites may occur to some degree, but it did not preclude the adoption of nest boxes by most breeding kestrels. We also found additional indications that many juvenile kestrels fledged from nest boxes emigrated to the surrounding natural areas.