To explore the patient-ventilator interaction of neurally adjusted ventilatory assist (NAVA) in patients with severe neurological diseases. A prospective study was conducted. Sixteen severe neurological patients with tracheotomy admitted to neurosurgery intensive care unit (NSICU) of Yijishan Hospital of the First Affiliated Hospital of Wannan Medical College from September 2019 to February 2020 were enrolled. According to the random number table method, they were treated with pressure support ventilation (PSV) mode followed by NAVA mode or NAVA mode followed by PSV mode mechanical ventilation. Each mode was ventilated for 24 hours. The number of auto-triggering, ineffective trigger, double trigger, inspiratory trigger delay, premature cycling, late cycling, and patient-ventilator asynchronous time (inspiratory trigger delay time, premature cycling time, and late cycling time) within 1 minute were recorded every 8 hours for 3 minutes. The average number of asynchronies per minute, asynchrony index (AI), total AI, asynchrony time, arterial blood gas analysis, and coefficient variation (CV%) of respiratory mechanics parameters of each asynchrony type between the two modes were compared. There were significant decrease in the number or AI of auto-triggering, ineffective trigger, inspiratory trigger delay, premature cycling, and late cycling with NAVA mode ventilation compared with PSV mode ventilation [auto-triggering times (times/min): 0.00 (0.00, 0.00) vs. 0.00 (0.00, 0.58), auto-triggering AI: 0.00 (0.00, 0.00) vs. 0.00 (0.00, 0.02), ineffective trigger times (times/min): 0.00 (0.00, 0.33) vs. 1.00 (0.33, 2.17), ineffective trigger AI: 0.00 (0.00, 0.02) vs. 0.05 (0.02, 0.09), inspiratory trigger delay times (times/min): 0.00 (0.00, 0.58) vs. 0.67 (0.33, 1.58), inspiratory trigger delay AI: 0.00 (0.00, 0.02) vs. 0.05 (0.02, 0.09), premature cycling times (times/min): 0.00 (0.00, 0.33) vs. 0.33 (0.08, 1.00), premature cycling AI: 0.00 (0.00, 0.01) vs. 0.02 (0.00, 0.05), late cycling times (times/min): 0.00 (0.00, 0.00) vs. 1.17 (0.00, 4.83), late cycling AI: 0.00 (0.00, 0.00) vs. 0.07 (0.00, 0.25), all P < 0.05]. But there was significant increase in the number or AI of double trigger with NAVA mode ventilation as compared with PSV mode ventilation [times (times/min): 1.00 (0.33, 2.00) vs. 0.00 (0.00, 0.00), AI: 0.04 (0.02, 0.11) vs. 0.00 (0.00, 0.00), both P < 0.05]. Total AI and incidence of total AI > 0.1 showed significant decrease during NAVA mode ventilation as compared with PSV mode ventilation [total AI: 0.08 (0.04, 0.14) vs. 0.22 (0.18, 0.46), incidence of total AI > 0.1: 37.50% (6/16) vs. 93.75% (15/16), both P < 0.01]. There was no significant difference in asynchronous time or arterial blood gas analysis between the two modes. There were significant increases in variances of peak airway pressure (Ppeak) and expiratory tidal volume (VTe) during NAVA mode ventilation as compared with PSV mode ventilation [Ppeak coefficient of variation (CV%): 11.25 (7.12, 15.17)% vs. 0.00 (0.00, 2.82)%, VTe CV%: (8.93±5.53)% vs. (4.71±2.61)%, both P < 0.05]. Compared with PSV mode, NAVA mode can reduce the occurrence of patient-ventilator asynchronous events, reduce the AI and the occurrence of serious patient-ventilator asynchronous events, so as to improve the patient-ventilator interaction. NAVA and PSV modes can achieve the same gas exchange effect. At the same time, NAVA mode has potential advantages in avoiding insufficient or excessive ventilation support, diaphragm protection and prevention of ventilator-induced lung injury.