AIM: This study aims to assess the acceptance of a VR-based disaster emergency nursing escape room teaching method among nurses and midwives and to explore the main factors influencing their acceptance. BACKGROUND: The increasing frequency of natural disasters due to global climate change poses a significant threat to human health. Effective training for nurses and midwives is critical as they are frontline responders in disaster relief. DESIGN: A quasi-experimental pretest-posttest design was used with a non-random sample to assess changes in acceptance and identify influencing factors. The study followed the TREND guidelines. METHODS: Questionnaires (n = 281) were distributed across four types of medical centers in China (obstetrics, public health, pediatrics, orthopedics) using targeted sampling. After analyzing valid responses (n = 247), participants with scores below the average (3.67) were selected (n = 160) for in-depth interviews. Data were analyzed using NVivo 11.0. A novel quasi-experimental project was developed that pioneered the integration of AI-driven narrative design (ChatGPT) and immersive VR production tools (Insta360 X3, Blender, After Effects) to create a disaster-themed escape room, deployed via an interactive online platform (Bilibili).Following the intervention, participants completed the System Usability Scale (SUS) and a post-intervention acceptance questionnaire. RESULTS: The intervention significantly increased participants' acceptance, with scores rising from 3.24 to 4.29, highlighting both its educational value and its feasibility for real-world training integration. CONCLUSION: The AI+VR-based escape room teaching method enhances disaster response capabilities and offers a new direction for nursing education. Further optimization of content and technical support will improve practical application.