Background: Fibrinogen is increasingly studied as an inflammatory biomarker in chronic obstructive pulmonary disease (COPD), and there are limited data about the role of fibrinogen to assess the severity of acute exacerbation of COPD. This study aimed to explore whether circulating fibrinogen could be used as a surrogate to measure the severity and predict the prognosis of acute exacerbation of chronic obstructive pulmonary disease (AECOPD).Methods: A total of 523 AECOPD patients diagnosed at our center from January 2016 to June 2021 were retrospectively enrolled in this study. Electronic medical record of each patient was retrieved to collect data regarding baseline characteristics and laboratory parameters, as well as the use of noninvasive positive pressure ventilation (NPPV) and patients’ prognosis. Multiple linear regression analyses were used to identify the independent factors that associated with fibrinogen values. Receiver-operating characteristic (ROC) curve and multivariate logistic regression analysis were applied to further verify the use of fibrinogen to predict NPPV failure. Results: Compared to patients with low levels of fibrinogen (≤3.5g/L), patients with increased fibrinogen levels (>3.5g/L) were associated with increased CRP expression, leukocyte, neutrophil counts and more frequent antibiotics use. In addition, the average fibrinogen level among patients with NPPV failure was significantly increased compared to non-NPPV patients and NPPV success patients. The presence of emphysema, pneumonia, history of long-term oxygen therapy (LTOT) and the CRP value were independent risk factors associated with increased fibrinogen levels in AECOPD. Furthermore, our data indicated that fibrinogen could be considered as an reliable biomarker to predict NPPV failure (AUC, 0.890; 95% CI 0.832–0.947) with an odds ratio of 6.1 (95% CI, 1.86-19.98; P=0.03). Conclusions: The level of fibrinogen could be used as a surrogate to measure severity, and among AECOPD patients who required NPPV, higher fibrinogen could independently predict NPPV failure.