Schools provide a favorable setting for health education, however, the most effective school-based exercise mode for improving physical fitness remains unclear. This network meta-analysis was designed to assess and rank the comparative efficacy of six exercise modalities on physical fitness indicators in a school-based setting. An online search of the Web of Science, PubMed, SPORTDiscus, and Scopus databases was conducted. Randomized and quasi-randomized controlled trials were considered. Outcomes included measures of anthropometry and body composition, muscular fitness, and cardiorespiratory fitness. Data were pooled with a random effects model using the frequentist framework. A total of 66 studies with 8,578 participants (48% girls) were included. High-intensity interval training was the most effective intervention reducing body mass index (mean difference (MD) = -0.60 kg·m-2, 95% confidence interval (95%CI) = -1.04 to -0.15, p = 0.009), elevating VO2max (MD = 3.59 mL·kg-1·min-1, 95% CI = 2.45 to 4.74, p < 0.001), and 20-meter sprint performance (MD = -0.35 s, 95% CI = -0.55 to -0.14, p = 0.001). Aerobic training had the highest probability of reducing waist circumference (standardized mean difference (SMD) = -0.60, 95% CI = -0.88 to -0.32, p < 0.001). Active video games emerged as a promising modality for improving countermovement jump (MD = 2.43 cm, 95% CI = 0.06 to 4.80, p = 0.041) and shuttle running performance (SMD = 0.86, 95% CI = 0.29 to 1.43, p = 0.003). Strength training was the best exercise mode for improving standing long jump performance (SMD = 1.03, 95% CI = 0.07 to 1.98, p = 0.035) while combined training was rated the first for decreasing body fat percent (MD = -2.56%, 95% CI = -4.73 to -0.40, p = 0.022) and increasing push-up repetitions (SMD = 3.59, 95% CI = 0.81 to 6.37, p = 0.012). School-based exercise interventions have multiple effects on physical fitness. The findings of this study will help to inform physical education teachers and coaches how best to deliver exercise programs in a school setting. Since the study was limited by the original research, the conclusions will require further verification using high-quality randomized controlled trials. PROSPERO, Identifier: CRD42023401963.