Objective To systematically evaluate the risk prediction models for secondary neurological complications in patients undergoing extracorporeal cardiopulmonary resuscitation （ECPR）during hospitalization, and to provide a basis for clinical practice.Methods A comprehensive search was conducted in databases including CNKI, Wanfang Data, CBM, PubMed, Web of Science, Embase, CINAHL, and Cochrane Library for studies on risk prediction models for secondary neurological complications in ECPR patients, covering the period from database inception to April 8, 2024.Two researchers independently screened the literature, assessed the quality of the studies, and extracted data.Results A total of 17 studies were included, involving 18 models with area under the receiver operating characteristic curve （AUC） ranging from 0.700 to 0.945.The top five repeatedly reported predictive factors of the models were age, pre-ECMO lactate level, shockable rhythm, time from CPR to ECMO initiation, and pre-ECMO pH value.There was ongoing debate regarding the time of lactate level measurement.All 17 included studies were deemed to have a high risk of bias, primarily due to inappropriate data sources, events per variable being less than 20, lack of model performance evaluation, and overfitting of models and so on; the overall applicability of the 14 studies were relatively good.Conclusion The risk prediction models for secondary neurological complications in ECPR patients during hospitalization are still imperfect.Future research and model development should strive to minimize bias and focus on the practicality and operability of models to provide scientific evidence for clinical decision-making, thereby optimizing patient treatment plans and improving prognosis.