<p>This cohort study investigated the possible association between serum zinc (SZn) concentration and the risk of progression to type 2 diabetes (T2D) in subjects with isolated impaired glucose tolerance (iIGT). SZn was measured in 198 subjects with iIGT (mean age: 53.0 ± 14.4 years and 33.8% were men) at baseline (2009–2011), and they were followed for developing T2D up to 2017. A univariate unrestricted regression spline (UVRS) was used to assess the potential non-linear association and identify the best placement of SZn knots related to the incidence of T2D. Multivariable Cox proportional hazard models were used to calculate adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) of incident T2D across the best placement of knots. The predictive power of SZn for developing T2D was determined using receiver operating characteristic (ROC) analyses, and the Youden index identified the optimal cut-off values. Mean baseline SZn concentration was 115 ± 42.9 µg/dL. Over a median 6-year follow-up, 27.8% of subjects with iIGT developed T2D. A non-linear association was observed between SZn and the incidence of T2D (<i>p</i>_{non-linearity} from the likelihood ratio test &lt; 0.001). Higher SZn ≥ 106 µg/dL and ≥ 134 µg/dL was associated with elevated risk of T2D by 127% (HR = 2.27, 95% CI = 1.01–5.10) and 144% (HR = 2.44, 95% CI = 1.05–5.69), respectively. According to the crude and multivariate-adjusted ROC analyses, the optimal cut-off values of SZn to identify incident T2D were ≥ 87.5 µg/dL (sensitivity of 83.6%, specificity of 30.1%) and 151.8 µg/dL (sensitivity of 64.4%, specificity of 76.3%), respectively. Elevated SZn levels at baseline were positively associated with the future risk of T2D in subjects with iIGT.</p>