OBJECTIVE To compare two levels of continuous cisatracurium-induced curarization in hypoxemic, ventilated patients. METHODS An open-labeled, multicenter, prospective, randomized study. METHODS Hundred two patients with a ratio between arterial oxygen tension and inspired oxygen tension (PaO(2)/FIO(2)) less than 200 despite optimization of sedation and ventilation were randomized into group 1 (n=52) with an end point of no response at orbicularis oculi to train-of-four (TOF) stimulation or group 2 (n=50) with an end point of two responses. RESULTS The PaO(2)/FIO(2) and end-inspiratory plateau airway pressure (Pplat) were evaluated at baseline (before curarization) and at regular intervals once TOF end points had been attained for up to 2 h afterwards (T2 h). A decrease of 1 cmH(2)O or more of Pplat at T2 h compared to baseline was observed in 37% and 50% of the patients in groups 1 and 2, respectively (p=0.17). Time courses of PaO(2)/FIO(2) (mmHg) and Pplat (cmH(2)O) [mean (SD)] were equivalent in both groups, with a mild increase in PaO(2)/FIO(2) [p=0.0014; from 126 (33) to 141 (55) and from 134 (40) to 152 (52), respectively, in groups 1 and 2] and decrease in Pplat [p=0.016; from 29.1 (8.9) to 28.5 (8.8) and from 27.7 (7.5) to 26.6 (7.6)]. Median total durations of curarization were 28.9 h (3.1-219.7) in group 1 and 31.4 h (1.6-650.6) in group 2. Median cisatracurium infusion rates were 5.2 microg kg(-1) min(-1) (2.1-13.7) in group 1 and 3.6 microg kg(1) min(-1) (1.0-13.5) in group 2. The median delay to recovery from paralysis was shorter in group 2 (0.75 h vs 1.25 h; p=0.0008). CONCLUSIONS When a prolonged curarization is decided upon in an ICU patient, a blockade at 2/4 at TOF at orbicularis oculi has similar effects on respiratory parameters as a blockade at 0/4, allowing a decrease in total administered doses and a shortening of the recovery of muscle strength after cessation of infusion.