Models of stress and health suggest that emotions mediate the effects of stress on health; yet meta-analytic reviews have not confirmed these relationships. Categorizations of emotions along broad dimensions such as valence (e.g., positive and negative affect) may obscure important information about the effects of specific emotions on physiology. Within the context of the integrated specificity model, we present a novel theoretical framework that posits that specific emotional responses associated with specific types of environmental demands influence cortisol and immune outcomes in a manner that would have likely promoted the survival of our ancestors. We analyzed experiments from 66 journal articles that directly manipulated social stress or emotions and measured subsequent cortisol or immune responses. Judges rated experiments for the extent to which participants would experience theoretically relevant cognition and affect clustered around five categories: (a) cognitive appraisals, (b) basic emotions, (c) rumination and worry, (d) social threat, and (e) global mood states. As expected, global mood states were unassociated with the effect sizes, whereas exemplars from the other categories were generally associated with effect sizes in the expected manner. The present research suggests that coping strategies that alter appraisals and emotional responses may improve long-term health outcomes. This might be especially relevant for stressors that are acute or imminent, threaten one's social status, or require extended effort. (Contains 9 figures, 6 tables and 6 footnotes.)