Objective To conduct a meta-analysis of risk prediction models for non-suicidal selfinjury（ NSSI） among adolescents， so as to provide scientific evidence for selecting appropriate models. Methods Relevant studies were systematically retrieved from PubMed， CINAHL， Web of Science， PsycINFO， Embase， SinoMed， The Cochrane Library， China National Knowledge Infrastructure， VIP， and WanFang Data. The search period was from database establishment to February 20， 2025. Two researchers independently conducted literature screening and data extraction. Model quality was assessed using the prediction model risk of bias assessment tool（ PROBAST）. Meta-analysis of the area under the curve for predictive models was performed using MedCalc 22.016. Meta-analysis of co-predictive factors was conducted using RevMan 5.4. Results A total of 13 studies were included. The area under the curve for the predictive model ranged from 0.74 to 0.976. The meta-analysis showed that the area under the curve of the pooled model was 0.859［ 95%CI（ 0.848， 0.869）］. Female［ OR=0.75， 95%CI（ 0.66， 0.85）］， depression severity［ OR=0.55， 95%CI（ 0.16， 1.87）］， childhood trauma［ OR=2.93， 95%CI（ 1.17， 2.21）］， and sleep disorders［ OR=1.44， 95%CI（ 0.84， 2.46）］ were predictive factors for adolescent NSSI. Conclusions Risk predictive models for adolescent NSSI are still in the developmental stage. It is recommended that the existing model be further optimized in the future.