Background: Many studies have attempted to link variation in corticosterone (Cort), the primary glucocorticoid (GC) in birds, with reproductive output, however the consequences of variation in Cort levels on survival probability have been relatively less explored. When considering baseline Cort and survival probability, two hypotheses have been proposed, the cort-fitness hypothesis and the cort-activity hypothesis. Notwithstanding a few exceptions, studies have generally however shown no relationship between levels of baseline Cort and the probability of survival. Methods: With the aim of increasing our understanding of the effect of baseline Cort on survival probability we utilized mark-recapture data from a long-term study (eight years of capture-mark-recapture) of the Thorn-tailed Rayadito ( Aphrastura spinicauda ), fitting different survival models with and without baseline Cort levels as covariates. Secondly, because the two populations in our study have different values of baseline Cort, we evaluated whether the association between baseline Cort and apparent inter-annual survival probability varied in relation to these two populations. Results: In the high latitude population (the population with lower baseline Cort levels), we observed a quadratic (inverted U-shape) relationship between baseline Cort levels and inter-annual survival probability; in the low latitude population, baseline Cort had no apparent effect on inter-annual survival probability. Conclusions: Because of the quadratic relationship, a relationship that only can be observed with long-term data, our approach emphasizes the importance of using a capture-recapture model. In addition, because the effect of baseline Cort on survival probability was found to be context-dependent, our work also highlights the importance of comparing different populations.