Objective: To explore the screening effect of obesity assessed by body fat indicators on persistent dyslipidemia among children. Methods: Data were obtained from the baseline and follow-up survey of 'School-based Cardiovascular and Bone Health Promotion Program.' BMI, fat mass index (FMI), and fat mass percentage (FMP) were used to define obesity. Dyslipidemia, diagnosed both in the baseline and a follow-up survey, was defined as persistent dyslipidemia. The area under the receiver operating characteristic curve (AUC) was used to compare the predictive capabilities of obesity defined by different indicators on persistent dyslipidemia. Results: A total of 10 783 children (boys accounted for 49.6%) were included in the analysis, with the average age as (10.9±3.3) years old. The detection rates of persistent high TC, high LDL-C, low HDL-C, high TG, and high non-HDL-C were 1.3%, 1.2%, 4.3%, 1.3%, and 0.8%, respectively. In boys, the capabilities of FMI- and FMP-defined obesity in the prediction of persistent high LDL-C [FMI: AUC=0.626 (95%CI: 0.558-0.694), P=0.024; FMP: AUC=0.642 (95%CI: 0.574-0.710), P=0.004] and high non-HDL-C [FMI: AUC=0.637 (95%CI: 0.584-0.689), P=0.017; FMP: AUC=0.641 (95%CI: 0.588- 0.693), P=0.018] were significantly higher than BMI-defined obesity. Besides, obese boys defined by FMI had the stronger capability in predicting persistent low HDL-C than that defined by BMI [AUC=0.784 (95%CI: 0.742-0.826) vs. 0.750 (95%CI: 0.726-0.773), P=0.047]. In girls, the capabilities of FMI- and FMP-defined obesity in the prediction of persistent dyslipidemia were not statistically different from BMI. Conclusions: The obesity assessed by body fat performed better in predicting persistent high LDL-C, low HDL-C, and high non-HDL-C than that assessed by BMI among boys, which can be further applied to cardiovascular disease prevention.