BACKGROUND: Population pharmacokinetic (popPK) models are essential tools for optimizing ustekinumab (UST) dosing for the treatment of inflammatory bowel disease (IBD) through therapeutic drug monitoring. The external validation of these models is necessary to ensure their predictive performance and clinical utility. The aim of the study was to externally validate 4 published popPK models of UST in a real-world cohort of patients with IBD using prediction-based and simulation-based diagnostics, as well as Bayesian forecasting. METHODS: Four popPK models of UST, identified through a systematic literature review, were evaluated using data from 99 patients with IBD and 374 serum UST concentrations. Predictive performance and Bayesian forecasting were assessed using statistical metrics, including mean prediction error, median prediction error (MDPE), and median absolute prediction error (MADPE). The acceptability criteria (MDPE +/-20%, MADPE </=30%, F20 >/=35%, and F30 >/=50%) were applied. RESULTS: None of the models satisfied the predefined acceptability criteria. The Xu et al model demonstrated the best performance, achieving an MDPE of 19.55% and the lowest RMSPE (2.88 mcg/mL), but F20 (20.1%) and F30 (32.4%) values fell below thresholds. The model proposed by Adedokun et al showed strong results in simulation-based diagnostics, with only 5.6% of the observed concentrations outside the prediction interval. CONCLUSIONS: The models developed by Xu et al and Adedokun et al exhibited the most promising predictive performance and potential clinical applicability for model-informed precision dosing. Refinements to these models and further research are required to enhance their use in personalized UST therapies for IBD.