Objective This is a meta-analysis of the efficacy of noninvasive ventilation (NIV) with helmet compared to NIV with face mask in patients with acute respiratory failure (ARF). Methods " Helmet, face mask or facial mask” and " mechanical ventilation or noninvasive ventilation” were used as key words both in Chinese and English to search all the trials in PubMed, OVID, Embase, Scopus and Cochrane Library, websites, reference lists of articles, CNKI and Wanfang Database from inception to December 2016. Two reviewers independently assessed the methodological quality of the trials and extracted information. Revman 5.3 was used for data analysis. Results Ten randomized controlled trials (RCTs) and six case-control trials were included. NIV with a helmet reduced the intubation rate (OR=0.35, 95%CI 0.24 to 0.51, P<0.000 01), in-hospital mortality rate (OR=0.51, 95%CI 0.34 to 0.76, P=0.001), and NIV-related complications (OR=0.10, 95%CI 0.06 to 0.15, P<0.000 01) compared to NIV with face mask. There was no significant difference in gas exchange between two groups. In the subgroup analysis, types of ARF and ventilation mode did not affect the intubation rate and the complications relevant to NIV, but NIV with helmet mainly decreased the in-hospital mortality of the patients with hypoxemic ARF or pressure support ventilation. Conclusions NIV with a helmet can decrease the endotracheal intubation rate, in-hospital mortality, and NIV-related complications of the patients with ARF. And helmet is as effective as face mask in improving the gas exchange. However, larger or multicenter RCTs are needed to analyze the role of NIV with a helmet in this condition.