Legal similar case retrieval is becoming increasingly important in the judicial field. Traditional methods for similar case retrieval largely rely on key-word matching, which fails to deeply understand the legal semantics of cases. To enhance retrieval accuracy and the ability to handle complex cases, we pro-pose legal similar case retrieval model based on concept tree and optimal transport (ConTree-OT CRM). This model constructs concept tree by hierarchically and modularly organizing case information, employs deep learning models (e.g., BERT) for semantic understanding, and calculates similarity using optimal transport distance. Through comparative experiments, it is demonstrated that the model outperforms traditional models, such as BM25 and TF-IDF, on multiple metrics. The experimental results show that the retrieval framework based on the concept tree not only improves the accuracy of case matching but also provides support for explainable legal decision-making. Future research will focus on further enhancing the model’s precision and capability to handle complex cases.