BACKGROUND: Women with gestational diabetes mellitus (GDM) are at high risk of developing hypertriglyceridemia (HTG) in late pregnancy, a condition associated with adverse maternal and neonatal outcomes. Early identification of high-risk individuals at GDM diagnosis is crucial for timely intervention. OBJECTIVE: To develop and internally validate a clinical risk prediction model for estimating the individualized risk of late-pregnancy HTG in women at the time of GDM diagnosis. METHODS: This single-center retrospective cohort study included 587 women with GDM. Predictor variables, available at or before GDM diagnosis (24-28 weeks), included demographic characteristics, glycemic markers from the 75g oral glucose tolerance test (OGTT), and first-trimester lipid profiles. The outcome was late-pregnancy HTG, defined as a triglyceride level ≥ 2.3 mmol/L. The dataset was randomly split into a training set (70%) and a held-out internal test set (30%) for secondary performance assessment, whereas bootstrap resampling of the full analytic cohort was used as the primary internal validation strategy to quantify optimism. Variable selection was performed using LASSO logistic regression with 10-fold cross-validation in the training set, followed by standard multivariable logistic regression. Model performance was assessed using discrimination, calibration-in-the-large, calibration slope, Brier score, scaled Brier score, and decision curve analysis. RESULTS: The incidence of late-pregnancy HTG was 32.7% (192/587). The final model incorporated five predictors: pre-gravid BMI, fasting plasma glucose (FPG) at diagnosis, 1-hour post-load glucose (1hPG) at diagnosis, first-trimester triglycerides (TG), and first-trimester high-density lipoprotein cholesterol (HDL-C). The model demonstrated good discrimination in the held-out internal test set, with an AUC of 0.816 (95% CI: 0.754-0.878), calibration-in-the-large of -0.03, calibration slope of 0.94, Brier score of 0.151, and scaled Brier score of 0.301. Bootstrap internal validation of the full cohort showed limited optimism, with an optimism-corrected AUC of 0.805, calibration-in-the-large of -0.02, calibration slope of 0.91, Brier score of 0.158, and scaled Brier score of 0.269. Decision curve analysis supported clinical utility across clinically relevant thresholds, and the full five-variable model provided modestly higher net benefit than the four-variable model excluding first-trimester TG. CONCLUSION: This study developed and internally validated a prediction model for late-pregnancy HTG in women with GDM using five routinely available clinical parameters. The model shows promising performance for individualized risk stratification at GDM diagnosis, potentially facilitating targeted monitoring and management. External validation in independent multicenter cohorts is required before clinical implementation.