A substantial portion of non-obese population is afflicted with Non-alcoholic Fatty Liver Disease (NAFLD). The Triglyceride Glucose (TyG) index, a quantifier of insulin resistance magnitude, is determined by the product of fasting plasma glucose and triglyceride concentrations. The relationship between the TyG index and NAFLD within this cohort remains ambiguous. We conducted a retrospective analysis utilizing datasets acquired from the Dryad digital repository. Non-obese participants (BMI < 25 kg/m2) were enrolled at the Wenzhou Medical Center of Wenzhou People's Hospital between January 2010 and December 2014. Demographic information and biochemical parameters were systematically compiled, and the diagnosis of NAFLD was established through ultrasonographic evidence. This study cohort included 16,172 non-obese participants with a 5-year follow-up, among whom 2,322 (14.36%) developed NAFLD. The disparity between TyG index quartiles in the accumulative incidence of new-onset NAFLD was distinct, with an increasing risk of new-onset NAFLD as the TyG index increased. Participants in highest quartile exhibited the maximum risk of NAFLD. In the fully adjusted model 3, the hazard ratios for NAFLD in Q2, Q3, and Q4 were 2.15 (1.62, 2.87), 2.89 (2.20, 3.80) and 4.58 (3.48, 6.02), respectively. Meanwhile, the TyG index and NAFLD risk showed a highly significant overall correlation (p < 0.0001) and nonlinearity (p < 0.0001) according to the limited cubic splines. In subgroup analysis, a significant interaction was noted between new-onset NAFLD and SBP (<140 mmHg vs. ≥140 mmHg; P for interaction = 0.0114). The SBP < 140 mmHg subgroup demonstrated an enhanced TyG index influence on NAFLD risk (HR = 2.83, 95% CI: 2.48-3.23, p < 0.0001). The TyG index serves as a straightforward instrument for assessing NAFLD risk in non-obese individuals, enabling prompt identification and management in this population segment.