<p>Similar Case Retrieval (SCR) is a core task in legal artificial intelligence, aiming to assist decision-making by identifying relevant precedent cases. In some existing works, legal elements are introduced to augment legal expertise, however, the performance is still limited due to focusing on a single aspect. Additionally, long legal documents are often truncated to meet the input limit of pre-trained language models (PLMs), making key information lost. To address these issues, we propose a key-Fact and multi-Legal-Element Joint Augmentation model, named FLEJA, to improve the accuracy and efficiency of SCR based on Chinese legal corpus. First, a key fact extraction module is designed to segment legal documents into sentences and extract those most semantically related to the judgment. It improves efficiency by reducing redundancy while preserving key information within PLM limit and providing refined text for legal element extraction. Second, a multi-legal- element extractor is introduced to obtain key legal elements, including charges, law articles, and prison terms, which is pre-trained on datasets of Legal Judgment Prediction (LJP) tasks. Finally, the multiple legal elements and key facts are integrated, enhancing SCR by assessing case similarity from multiple perspectives. We conduct extensive experiments on two public datasets. Compared to baseline models, our model is validated by about 10%, 12%, and 82% average improvement in terms of accuracy, mean average precision, and inference speed, respectively, which indicate that our model effectively utilizes multiple legal elements to enhance the performance of SCR task and its potential application in legal artificial intelligence.</p>