The state of nurses' work engagement affects their nursing quality and the level of medical services. Increasing the level of nurses' work engagement plays an important role in improving the clinical treatment effect as well as improving patient satisfaction. A good nurse-physician collaboration can effectively relieve the psychological pressure of nurses and improve the work efficiency of both sides. In addition, it can enhance the comprehensive understanding of patients, the course of a disease, and reduce the chance of making mistakes. Therefore, this study aims to explore the specific mechanism of nurses' risk perception and professional identity between nurse-physician collaboration and work engagement. Nurses from the primary, secondary, and tertiary hospitals in Shanghai were selected as research participants. This study was conducted from 11/2018 to 05/2019. A total of 1015 questionnaires were distributed, in which 977 questionnaires were valid by screening and eliminating the invalid questionnaires. The effective recovery rate was 96.3%. In this study, there were 84 males (8.6%) and 893 females (91.4%). This distribution of gender is consistent with that of the professional group of nurses. The age range of the participants was from 17 to 59 years old (M = 31, SD = 7). The level of nurse-physician collaboration, risk perception level, professional identity level, and work engagement were measured by the Nurse-Physician Collaboration Scale, Risk Perception Questionnaire for Nurses, Professional Identity Scale for Nurses, and the Utrecht Work Engagement Scale (UWES), respectively. The Cronbach α coefficients of the four questionnaires in this study were .969, .971, .965, and .962, respectively. Data were analyzed with SPSS 25.0 and Mplus 7.4. Additionally, the bias-corrected percentile Bootstrap method was used to analyze the chain mediating effect of risk perception and professional identity. Results showed there were no obvious common methods bias using the Harman single factor analysis. The results indicated that (1) by controlling the variables such as education level and professional title, there was a statistically significant positive relationship between nurse-physician collaboration and nurses' professional identity (r = .666, p < .001), nurse-physician collaboration and work engagement (r = .722, p < .001). Furthermore, there was a statistically significant positive relationship between professional identity and work engagement (r = .796, p < .001). In addition, there was a statistically significant negative relationship between risk perception and work engagement (r = −.448, p < .001). (2) The structural equation model revealed that the data fit the theoretical model well (χ2/df = 5.69, CFI = .961, TLI = .955, SRMR = .057, RMSEA = .069). (3) Risk perception and professional identity play a chain mediating role between nurse-physician collaboration and nurses’ work engagement. The chain mediating effect accounted for 4.38% of the total effect, 95% CI: [.023, .049]. The indirect effect of risk perception on work engagement accounted for 6.7% of the total effect, 95% CI: [.037, .068]; the indirect effect of professional identity on work engagement accounted for 44.20% of the total effect, 95% CI: [.293, .394]. These findings suggest reducing nurses' professional risk perception and enhancing their professional identity can effectively improve nurses’ work engagement. Therefore, to provide better nursing services for patients and alleviate the conflicts between medical workers and patients, it is important to provide adequate support for nurses in their daily work, maintain a good and safe working environment, reduce nurses’ professional risks, and strengthen their professional identity. In addition, the good working collaboration between doctors and nurses is a crucial basis for each other's effective work, and it is also an essential factor in improving the level of medical collaboration.