Objective Risk factors for chronic diseases such as insufficient physical activity (PA), overweight or hypertension are getting more predominant among children. While school-based interventions are promising preventive strategies, evidence of their long-term effectiveness, especially among vulnerable populations is scarce. KaziKidz is a physical education intervention that promotes healthy and active lifestyles among children from low-income communities. We aim at assessing the short- and long-term effects of KaziKidz on cardiovascular risk factors in high-risk children from disadvantaged communities. Methods The intervention was tested in a cluster-randomized controlled trial between January and October 2019 in eight primary schools in South Africa. Children with adverse cardiovascular profiles were defined as presenting at least one risk factor for (1) overweight, (2) elevated blood pressure, (3) glycaemia, and/or (4) cholesterolemia. After their identification, high-risk children were re-assessed 2-years post-intervention. Study outcomes include accelerometry-measured PA (MVPA), body-mass-index (BMI), mean arterial pressure (MAP), glucose (HbA1c), and lipid levels (TC/HDL). We conducted mixed regression analyses to assess intervention effects by cardiovascular risk profile, and Wilcoxon signed-rank tests to evaluate longitudinal changes in the high-risk subpopulation. Results A significant intervention effect on MVPA during school hours was found for physically inactive children (B: 1.71, 95% CI: 0.14 – 3.35, p-value: 0.008), and among active (B: 1.80, 95% CI: -0.22 – 3.82, p-value: 0.035) and inactive (B: 2.03, 95% CI: 0.58 – 3.42, p-value: < 0.001) girls. In contrast, the intervention lowered HbA1c (B: -0.26, 95% CI: -0.52 – -0.01, p-value: 0.037) and TC/HDL (B: -0.11, 95% CI: 0.18 – -0.05, p-value: 0.002) only in children with normal glucose or lipid values, respectively. The intervention effects were not maintained in at-risk children at follow-up. A decline was observed in MVPA from 53.29 to 41.90 min/day (z = -1.95, p = 0.05, r = 0.26), and an increase in BMI-for-age (z = -0.34, p = 0.737, r = 0.03), MAP (z = -5.18, p < 0.001, r = 0.42), HbA1c (z = -1.63, p = 0.104, r = 0.25) and TC/HDL (z = -1.61, p = 0.11, r = 0.21). Conclusion We conclude that schools are key settings to promote PA and improve health; however, because the intervention effects were not maintained in children at risk of NCDs, structural changes that ensure effective interventions reach disadvantaged populations are necessary to achieve sustainable impact.