<p>Acute kidney injury (AKI) is a concerned complication in patients with cancer receiving anti-PD-1/PD-L1 therapy, with severe AKI linked to adverse outcomes. Here, we constructed and validated a predictive model for severe AKI in these patients. Patients administered anti-PD-1/PD-L1 antibody at the Dongyang People’s Hospital from January 2019 to December 2023 (831patients) were included and randomly assigned into training and testing sets in a 7:3 ratio. An external validation dataset (907 patients) was obtained from Zhejiang Provincial People’s Hospital. Severe AKI was defined AKI stages 2 and 3, based on the kidney disease improving global outcomes criteria. Severe AKI occurred in 4.2% (73/1738) of patients: 5.3% (31/581), 5.2% (13/250), and 3.2% (29/907) in the training, testing, and external validation sets, respectively. Overall survival was significantly lower in patients with severe AKI. In the training set, a nomogram for severe AKI was constructed using three predictive factors: higher systolic blood pressure, lower serum albumin level, and diuretic use. In the training set, the model achieved C-indices of 0.850, 0.829, and 0.808 for predicting severe AKI at 90, 180, and 360 days, respectively. Corresponding C-indices in the test set were 0.710, 0.775, and 0.857, while those in the external validation cohort reached 0.720, 0.772, and 0.757, demonstrating strong discriminability. Calibration charts and decision curve analyses confirmed its calibration capability and clinical utility. The developed nomogram aids in predicting severe AKI risk in patients receiving anti-PD-1/PD-L1 antibodies and supports effective preventive interventions.</p>