First, it is surprising that significant differences in the evaluated outcomes are found in studies with such small sample sizes (<70 patients per study in all studies). Second, all the studies except the trial by Trevisan et al6 were carried out in respiratory units admitting mostly patients with COPD. This is not the rule in most ICUs.7 Third, the control groups (invasive weaning) in these studies do not reflect current clinical practice. For instance, in the study most favorable to NIPPV,5 outcomes in patients assigned to invasive weaning were: rate of nosocomial pneumonia, 59%; rate of reintubation, 21%; need for tracheostomy, 59%; and ICU mortality, 41%. To try to compare those data with the real world, we have searched in the databases of two international studies on mechanical ventilation.7 In these databases, we have selected patients with COPD who required mechanical ventilation for >3 days and had a duration of weaning >3 days. This patient population would be similar to that in the study by Ferrer et al.5 From a total cohort of 10,151 patients who were mechanically ventilated, we found 160 patients meeting the above-mentioned criteria. The outcomes of this cohort were: rate of nosocomial pneumonia, 7%; rate of reintubation, 21%; need for tracheostomy, 9%; and ICU mortality, 10% (data not previously published). Last, the withdrawal of the endotracheal tube for patients failing a spontaneous breathing trial could raise ethical issues.