OBJECTIVE To compare the efficacy of noninvasive positive pressure ventilation (NPPV) by helmet and face mask in immunocompromised patients with hypoxemic acute respiratory failure (ARF). METHODS Case-control study. METHODS The general ICU of a university hospital. METHODS Nineteen immunocompromised patients (hematologic malignancies [n = 8], solid-organ recipients [n = 8], AIDS [n = 3]) with hypoxemic ARF, fever, and lung infiltrates were treated with NPPV delivered by a helmet. Nineteen immunocompromised patients matched for diagnosis, age, simplified acute physiology score II, and Pao(2)/fraction of inspired oxygen (Fio(2)) receiving NPPV through a facial mask served as case-control subjects. RESULTS The use of NPPV delivered via helmet was as effective as NPPV delivered via face mask in avoiding endotracheal intubations (intubation rate, 37% vs 47%, respectively; p = 0.37) and improving gas exchange; 14 patients (74%) in the helmet group showed a sustained improvement in Pao(2)/Fio(2) ratio (ability to increase Pao(2)/Fio(2) ratio > 200, or an increase > 100 from the baseline) in comparison with 7 patients (34%) in the mask group (p = 0.02), whose Pao(2)/Fio(2) at treatment discontinuation was higher (p = 0.02) and had fewer complications related to NPPV (ie, skin necrosis, p = 0.01). Moreover, the patients receiving ventilation via helmet required significantly less NPPV discontinuations in the first 24 h of application (p < 0.001) than patients receiving ventilation via face mask. CONCLUSIONS The helmet may represent a valid alternative to a face mask in immunocompromised patients with lung infiltrates and hypoxemic ARF, increasing the patient's tolerance (ie, the number of hours of continuous NPPV use without interruptions) and decreasing the rate of complications directly related to the administration of NPPV.