Background Kidney biopsy is an essential procedure for diagnosing kidney diseases. However, bleeding remains a major complication. This study aimed to evaluate predictive factors, including platelet aggregation testing and thromboelastography, and to develop a prediction model to estimate the probability of bleeding complications in patients undergoing kidney biopsy.Methods In this prospective cohort study, adult patients (aged ≥ 18 years) with renal failure were enrolled. Platelet aggregation testing and thromboelastography were performed within 24 h prior to kidney biopsy. A prediction model was developed using backward stepwise regression. Model performance was assessed using discrimination and calibration, and internal validation was performed using bootstrap resampling.Results A total of 119 patients were included. Bleeding complications occurred in 66 patients. Neutrophil-to-lymphocyte ratio ≥ 15, fibrinogen ≥ 500 mg/dL, time to maximum aggregation of epinephrine ≥ 450 s, and time to maximum aggregation of ristocetin ≥ 420 s were associated with a reduced risk of bleeding complications. In contrast, maximum lysis at 60 min (ML60) in extrinsic pathway thromboelastometry (EXTEM) ≥ 5% was associated with an increased risk of bleeding. The BENREEF model demonstrated good performance, with strong discrimination and calibration, including in internal validation.Conclusions Both protective and risk factors for bleeding complications following kidney biopsy were identified. The BENREEF model showed good predictive performance for estimating bleeding risk. However, external validation is required before clinical application.