Well-developed theories are necessary to guide the public in increasing physical activity (PA) and promoting physical health. The role of positive affective variables (PAVs) in exercise is gaining more attention, but none of the literature has provided a systematic review and quantitative analysis of its mediating role. Therefore, the purposes of this study are (1) to systematically review studies of PA interventions, that use PAVs as the mediating variables, in order to evaluate and provide narrative summaries of these studies; (2) to statistically synthesize evidence for the mechanism of the effects of PAVs on PA outcomes. To conduct an extensive search, a PRISMA-compliant protocol was completed, and five electronic databases had been searched by 1 April 2020. We used a two-stage structural equation modeling (TSSEM) analysis approach to test how interventions trigger the critical PA change process to influence outcomes. The search strategy generated 1,732 papers potentially relevant to this study; forty of these studies met the data extraction criteria for meta-analytic mediation analysis. The path coefficient from intervention to PAV = 0.26 (95% CI = 0.08 to 0.44), the path coefficient from PAV to PA = 0.21 (95% CI = 0.13 to 0.28), and the direct effect from intervention to PA is also significant ( = 0.19, 95% CI = 0.12 to 0.26). In addition, the indirect effect of intervention on PA via PAV was statistically significant (' = 0.05, 95% CI = 0.02 to 0.10). This reveals that PAVs partially mediate the relationship between interventions and PA. Our study is the first to systematically summarize the effects of experimental studies to increase PA through PAVs. It is highly recommended to make future interventions more innovative and to target the PAVs as mediators with higher fidelity.