To compare the effects of noninvasive assist-control ventilation (ACV) and pressure support ventilation (PSV) by nasal mask on respiratory physiologic parameters and comfort in acute hypercapnic respiratory failure (AHRF). A prospective randomized study. A medical ICU. Fifteen patients with COPD and AHRF were consecutively and randomly assigned to two noninvasive ventilation (NIV) sequences with ACV and PSV mode, spontaneous breathing (SB) via nasal mask being used as control. ACV and PSV settings were always subsequently adjusted according to patient's tolerance and air leaks. Fraction of inspired oxygen did not change between the sequences. ACV and PSV mode strongly decreased the inspiratory effort in comparison with SB. The total inspiratory work of breathing (WOBinsp) expressed as WOBinsp/tidal volume (VT) and WOBinsp/respiratory rate (RR), the pressure time product (PTP), and esophageal pressure variations (deltaPes) were the most discriminant parameters (p<0.001). ACV most reduced WOBinsp/VT (p<0.05), deltaPes (p<0.05), and PTP (0.01) compared with PSV mode. The surface diaphragmatic electromyogram activity was also decreased >32% as compared with control values (p<0.01), with no difference between the two modes. Simultaneously, NIV significantly improved breathing pattern (p<0.01) with no difference between ACV and PSV for VT, RR, minute ventilation, and total cycle duration. As compared to SB, respiratory acidosis was similarly improved by both modes. The respiratory comfort assessed by visual analog scale was less with ACV (57.23+/-30.12 mm) than with SB (75.15+/-18.25 mm) (p<0.05) and PSV mode (81.62+/-25.2 mm) (p<0.01) in our patients. During NIV for AHRF using settings adapted to patient's clinical tolerance and mask air leaks, both ACV and PSV mode provide respiratory muscle rest and similarly improve breathing pattern and gas exchange. However, these physiologic effects are achieved with a lower inspiratory workload but at the expense of a higher respiratory discomfort with ACV than with PSV mode.