Glucocorticoids are a popular tool for monitoring health of animal populations because they can increase with environmental stressors and can indicate chronic stress. However, individual responses to stressors create variation in the glucocorticoid–fitness relationship within populations. The inconsistency in this relationship calls into question the widespread use of glucocorticoids in conservation. We investigated the sources of variation in the glucocorticoid–fitness relationship by conducting a meta-analysis across a diverse set of species exposed to conservation-relevant stressors. We first quantified the extent to which studies inferred population health from glucocorticoids without first validating the glucocorticoid–fitness relationship in their own populations. We also tested whether population-level information like life history stage, sex and species longevity influenced the relationship between glucocorticoids and fitness. Finally, we tested for a universally consistent relationship between glucocorticoids and fitness across studies. We found more than half of peer-reviewed studies published between 2008 and 2022 inferred population health solely based on glucocorticoid levels. While life history stage explained some variation in the relationship between glucocorticoids and fitness, we found no consistent relationship between them. Much of the variation in the relationship could be the result of idiosyncratic characteristics of declining populations, such as unstable demographic structure, that coincided with large amounts of variation in glucocorticoid production. We suggest that conservation biologists capitalize on this variation in glucocorticoid production by declining populations by using the variance in glucocorticoid production as an early warning for declines in population health.