According to the cort-fitness hypothesis, glucocorticoid levels correlate negatively with fitness. However, field studies found mixed support for this hypothesis, potentially because the association between glucocorticoids and fitness might depend on prevailing environmental conditions. Based on the long-term monitoring of a natural rodent population, we tested whether individuals with elevated corticosterone levels were more likely to disappear, accounting for individual condition and among-year variation in food availability, population density and predation pressure. We used basal corticosterone levels measured at the onset of the pre-breeding season in 331 African striped mice from six generations. While basal corticosterone levels were highly repeatable within individuals, between-individual variation was large. Survival analysis revealed that disappearance risk over the pre-breeding season increased with elevated basal corticosterone levels for light but not for heavy individuals. High levels of corticosterone may be more deleterious to smaller individuals (i.e. through allostatic overload), eventually increasing their mortality risk, and disappearance would represent actual death. An alternative non-exclusive explanation could be that high levels of corticosterone selectively trigger dispersal in light individuals, and disappearance would rather reflect their departure from the population. Although environmental conditions varied considerably among generations, none of the interactions between corticosterone and environmental variables were significant. Disappearance probability was positively correlated with both predation pressure and with food availability, a factor favoring dispersal. In sum, elevated basal corticosterone levels increased disappearance in light striped mice, either directly via reduced survival prospects and/or indirectly via dispersal.