Screening for non-alcoholic fatty liver disease (NAFLD) is key step for primary management of fatty liver in the clinical setting. Excess weight subjects carry a greater metabolic risk even before exhibiting pathological patterns, including diabetes. We characterized the cross-sectional relationship between routine circulating biomarkers and NAFLD in a large sample of diabetes-free subjects with overweight or obesity, to elucidate any independent relationship. A population sample of 1232 consecutive subjects with a body mass index of at least 25 kg/m2, not receiving any drug or supplemental therapy, was studied. Clinical data and routine biochemistry were analyzed. NAFLD was defined using the validated fatty liver index (FLI), classifying subjects with a score ≥ 60% as at high risk. Due to extreme skewing of variables of interest, resampling matching for age and sex was performed. Our study population was characterized by a majority of females (69.90%) and a prevalence of NAFLD in males (88.90%). As a first step, propensity score matching was explicitly performed to balance the two groups according to the FLI cut-off. Based on the resulting statistical trajectories, corroborated even after data matching, we built two logistic regression models on the matched population (N = 732) to verify any independent association. We found that each unit increase of FT3 implicated a 50% increased risk of NAFLD (OR 1.506, 95%CI 1.064 to 2.131). When including glycated haemoglobin (HbA1c) in the model, free-triiodothyronine (FT3) lost significance (OR 1.557, 95%CI 0.784 to 3.089) while each unit increase in HbA1c (%) indicated a significantly greater NAFLD risk, by almost two-fold (OR 2.32, 95%CI 1.193 to 4.512). Glucose metabolism dominates a key pathway along the hazard trajectories of NAFLD, turned out to be key biomarker in monitoring the risk of fatty liver in diabetes-free overweight subjects. Each unit increase in HbA1c (%) indicated a significantly greater NAFLD risk, by almost two-fold, in our study.