AIMSTo examine the relationship between lipid profile and non-alcoholic fatty liver (NAFL), compare the predictive strengths of different lipid indicators to NAFL, and explore the possible mechanisms.METHODSMale workers from a baseline survey of a cohort of workers in southern China were included. Basic information was collected through face-to-face interviews. Plasma concentrations of fasting plasma glucose (FPG), total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C) were determined using a blood biochemical analyzer. Liver sonography was used to identify NAFL cases. Regression models were used to calculate ORs, and examine the association between C-reactive protein (CRP) levels and lipid profiles. Restricted cubic spline regression with four knots was used to examine the dose-response relationship, and mediation analysis was employed to examine the mediation effect.RESULTSh Among the 4016 male workers, 829 (20.64%) were diagnosed with NAFL. Compared with normal lipid profile, individuals with abnormal lipid profile had higher prevalence of NAFL (OR=2.27, 95%CI: 1.85-2.79 for TG; OR=1.45, 95%CI: 1.03-2.04 for TC; OR=1.56, 95%CI: 1.21-2.02 for HDL; OR=1.65, 95%CI: 1.25-2.18 for LDL; OR=2.28, 95%CI: 1.87-2.77 for dyslipidaemia) after adjusting for potential confounders. Dose-response relationships were observed among TG, HDL, and NAFL. In addition, no significant mediation effect of C-reactive protein (CRP) was found in the association between lipid profiles and NAFL.CONCLUSIONSAbnormal TG, TC, HDL, and LDL levels were all positively associated with NAFL, while CRP has no mediating effect, and TG tended to be a better predictor of NAFL.