Objective: The health action process approach (HAPA) is a social–cognitive model specifying motivational and volitional determinants of health behavior. A meta-analysis of studies applying the HAPA in health behavior contexts was conducted to estimate the size and variability of correlations among model constructs, test model predictions, and test effects of past behavior and moderators (behavior type, sample type, measurement lag, study quality) on model relations. Method: A literature search identified 95 studies meeting inclusion criteria with 108 independent samples. Averaged corrected correlations among HAPA constructs and multivariate tests of model predictions were computed using conventional meta-analysis and meta-analytic structural equation modeling, with separate models estimated in each moderator group. Results: Action and maintenance self-efficacy and outcome expectancies had small-to-medium sized effects on health behavior, with effects of outcome expectancies and action self-efficacy mediated by intentions, and action and coping planning. Effects of risk perceptions and recovery self-efficacy were small by comparison. Past behavior attenuated the intention-behavior relationship. Few variations in model effects were observed across moderator groups. Effects of action self-efficacy on intentions and behavior were larger in studies on physical activity compared with studies on dietary behaviors, whereas effects of volitional self-efficacy on behavior were larger in studies on dietary behaviors. Conclusions: Findings highlight the importance of self-efficacy in predicting health behavior in motivational and volitional action phases. The analysis is expected to catalyze future research including experimental studies targeting change in individual HAPA constructs, and longitudinal research to examine change and reciprocal effects among constructs in the model.