Contal et al18 provide data supporting the contention that bilevel PAP with a timed backup rate is superior to the S mode and that lower BURR may enhance the comfort of patients with OHS, answering an interesting but narrowly drawn question. Given that the identical recommendation was made by consensus in the AASM guidelines (with BURR at or slightly less than the sleeping spontaneous rate),17 it is gratifying to see experimental confirmation in this small, randomized, prospective study. However, in every research study, as in life, “some rain must fall”; here, the precipitation is somewhat of a drizzle but must still be noted. First, one must question whether the 10 subjects were adequately characterized, because patients with one or the other subtype of OHS might respond differently to the bilevel PAP modes tested.6 Were these patients all “true Pickwickians” who required treatment with bilevel PAP, or did the group include some individuals who could have been CPAP responders? Might the latter patients have responded well to just the S mode? Second, many patients were clearly not on optimal settings of bilevel PAP-S at home. The at-home levels of inspiratory PAP and expiratory PAP were maintained for the three study trials, even though the obstructive AHI was elevated in virtually all the patients during the bilevel PAP-S trials (median, 22; IQR, 15.3-37.3) and also in many of the patients during the bilevel PAP-S/T nights. Particularly during bilevel PAP-S treatment, repeated arousals from obstructive events could have prompted central or mixed apneas upon returning to sleep. Moreover, frank obstructive apneas would have contributed to PVA because inspiratory efforts without airflow cannot trigger spontaneous bilevel PAP breaths. Supporting this conjecture, the median value for PVA was 22% of total sleep time (IQR, 7%-20%) for the bilevel PAP-S trials. Interestingly, expiratory PAP settings (average ± SD) were only 9.2±1.8 cm H2O, arguably somewhat low for adequate suppression of obstructive apneas in this morbidly obese population (mean BMI ± SD, 48.5±5.1 kg/m2) when compared with published experience.16,22 Finally (as acknowledged by the authors), the number of subjects studied is indeed small, although this is understandable considering the 3 nights of PSG required. Nonetheless, this may have resulted in type 2 error and failure to demonstrate significance in improving some of the parameters of sleep architecture; in particular, the microarousal index and proportion of slow-wave sleep seemed to exhibit a trend toward improvement on the bilevel PAP-S/T modes.