Objective <p>Physical activity interventions (PAIs) are often recommended to ameliorate childhood obesity, but the effectiveness of theory-based PAIs remains unclear. This study aimed to synthesize evidence and quantify the impact of theory-based PAIs on body mass index (BMI) and BMI <i>z</i>-score in this population. Literature published from inception to June 2025 were retrieved from five major databases (Medline, Scopus, Cochrane, Embase, and Web of Science). Risk of bias was evaluated using the Cochrane RoB 2 tool, and meta-analysis was conducted in RevMan 5.3 using random-effects models to estimate the combined effect size (standardized mean difference [SMD]). Subgroup analyses (by theory type, age, intervention duration, and family engagement) and two types of sensitivity analyses were performed to explore sources of heterogeneity and assess the robustness of the findings. A total of 24 studies were included, 20 of which employed a single theoretical framework and 4 that integrated multiple theories, with social cognitive theory being the most frequently applied. The overall methodological quality of the included studies was generally moderate. Meta-analysis showed a statistically significant but modest effect of theory-based PAIs on BMI outcomes (SMD = −0.12; 95% CI: −0.19 to −0.04; <i>p</i> = 0.002). Moreover, subgroup analyses revealed that younger participant age, shorter intervention duration, deeper family engagement roles, and application of cognitive behavioral theory may have more positive effects. These findings indicate that theory-based PAIs are associated with meaningful benefits in preventing and managing childhood obesity. Future research should build on these findings by systematically incorporating favorable intervention components and strengthening the integration and reporting of theoretical frameworks to enhance the health and well-being of children and adolescents.</p>