This paper provides a comprehensive systematic literature review of the application of large language models (LLMs) in cybersecurity, addressing three key research questions: the construction of cybersecurity-oriented LLMs, their applications in various cybersecurity tasks, and the existing challenges and future research directions. The review covers over 180 works, spanning 25 LLMs and more than 10 downstream scenarios. It highlights the potential benefits and challenges of LLMs in enhancing cybersecurity practices, such as vulnerability detection, secure code generation, program repair, and threat intelligence. The paper also discusses the methods for constructing domain-specific LLMs, including continual pre-training and supervised fine-tuning, and evaluates the cybersecurity capabilities of various LLMs using specific datasets. Additionally, it explores the practical applications of LLMs in cybersecurity, such as threat intelligence generation, code review, and automated program repair, and identifies emerging challenges and future research directions. The study aims to bridge the gap between LLM advancements and their potential impact on cybersecurity, providing valuable insights and practical guidance for researchers and practitioners.This paper provides a comprehensive systematic literature review of the application of large language models (LLMs) in cybersecurity, addressing three key research questions: the construction of cybersecurity-oriented LLMs, their applications in various cybersecurity tasks, and the existing challenges and future research directions. The review covers over 180 works, spanning 25 LLMs and more than 10 downstream scenarios. It highlights the potential benefits and challenges of LLMs in enhancing cybersecurity practices, such as vulnerability detection, secure code generation, program repair, and threat intelligence. The paper also discusses the methods for constructing domain-specific LLMs, including continual pre-training and supervised fine-tuning, and evaluates the cybersecurity capabilities of various LLMs using specific datasets. Additionally, it explores the practical applications of LLMs in cybersecurity, such as threat intelligence generation, code review, and automated program repair, and identifies emerging challenges and future research directions. The study aims to bridge the gap between LLM advancements and their potential impact on cybersecurity, providing valuable insights and practical guidance for researchers and practitioners.