Glucocorticoid (cort) hormones are increasingly applied in studies of free-ranging animals, with elevated baseline cort levels generally assumed to indicate individuals or populations in worse condition and with lower fitness (the Cort-Fitness Hypothesis). The relationship between cort and fitness is rarely validated and studies investigating the cort-fitness relationship often find results inconsistent with the Cort-Fitness Hypothesis. The inconsistency of these studies may result in part from variation in the cort-fitness relationship across life history stages. Here we address the following questions in a two-year study in free-ranging tree swallows (Tachycineta bicolor): (1) Do baseline cort levels correlate with fitness within a life history stage? (2) Does the cort-fitness relationship vary across different life history stages? (3) Does the cort-fitness relationship vary across life history stages within an individual? (4) Does reproductive effort influence cort levels, and do cort levels influence reproductive effort? We measured baseline cort and fitness components in female birds of known breeding stages. We find correlations between baseline cort levels and fitness within some life history stages, but the relationship shifts from negative during early breeding to positive during late breeding, even within the same individuals. A positive relationship between baseline cort and fitness components during the nestling period suggests that reproductive investment may elicit higher cort levels that feedback to reallocate more effort to reproduction during critical periods of nestling provisioning. Our findings provide reason to question the Cort-Fitness Hypothesis, and have implications for the application of cort measures in monitoring the condition of populations of conservation concern.