Objective: Two measures of affect-affective attitude (AA) and anticipated affective reaction (AAR)- have frequently been used individually, but rarely simultaneously, in correlational studies predicting health behaviors. This research assessed their individual and combined impact in predicting intention and action for a range of health behaviors, controlling for theory of planned behavior (TPB) variables. Method: Self-reported intentions and performance of health behaviors were the main outcome measures. Design: Study 1 is a meta-analysis of published studies (k = 16) measuring the relevant variables. In Study 2, adults (N = 426) completed questionnaires assessing TPB variables, past behavior, AA, AAR, and subsequent behavior for a range of health behaviors. Results: Across both studies, AA and AAR were only moderately intercorrelated, although both had significant correlations with both intentions and behavior. AA was a significant predictor of intentions and behavior after controlling for TPB variables (Studies 1 and 2) plus past behavior (Study 2). In Study 1, AAR was a significant predictor of behavior, but not intentions, when controlling for TPB variables. In Study 2, AAR was a significant predictor of intentions when controlling for both TPB variables plus past behavior (Study 2), but was not a significant predictor of behavior when controlling for either of these variables. Several relationships were moderated by health-behavior category. Conclusions: Both AA and AAR are important predictors of health behaviors and can have independent effects on intentions and action. Studies manipulating both variables to test their independent and combined effects on behavior change are required.