A machine learning (ML) model was previously developed to predict docetaxel-induced infusion-related reactions (IRRs) in cancer patients, identifying three key predictors: younger age (<50), combination therapy, and elevated albumin (>4.4 g/dL). This study aimed to validate this model using two independent cohorts. The XGBoost model, which performed best in development (area under the receiver operating characteristic curve (AUROC) 0.71, sensitivity 0.72), was validated. A retrospective cohort from Seoul National University Bundang Hospital was used for external validation, with performance evaluated by AUROC and sensitivity. Prospective validation was conducted in a cohort of docetaxel-treated patients in 2024 at Seoul National University Hospital, where sensitivity was assessed. In both cohorts, predicted probabilities were analyzed for patients who experienced IRRs. The model yielded an AUROC of 0.67 and sensitivity of 0.71 in the retrospective cohort, and a sensitivity of 0.71 in the prospective cohort. In both cohorts, the distribution of predicted probabilities among IRR cases showed a concentration in the higher probability ranges. An ML model predicting docetaxel-induced IRR was validated in external and prospective cohorts, demonstrating satisfactory performance with high sensitivity. Leveraging routinely available data, the model supports generalizability and effective risk stratification in clinical practice.