Self-report measures assess mental processes or representations that are consciously accessible. In contrast, implicit measures assess automatic processes that often operate outside awareness. Whereas self-report measures have often failed to show expected relationships with endocrine stress responses, little effort has been made to relate implicit measures to endocrine processes. The present work examines whether implicit affectivity as assessed by the Implicit Positive and Negative Affect Test (IPANAT) predicts cortisol regulation. In Study 1, implicit low positive affectivity, but not negative affectivity, significantly predicted circadian cortisol release. In Study 2, implicit negative affectivity, but not positive affectivity, significantly predicted the cortisol response to acute stress. By contrast, cortisol regulation was not predicted by self-reported affectivity. The findings support the use of implicit affectivity measures in studying individual differences in endocrine stress responses and point to a differential role of positive and negative affectivity in baseline versus stress-contingent cortisol release, respectively.