Intermittent positive pressure ventilation administered nocturnally via a nasal mask has been associated with improvements in pulmonary function and symptoms in patients with restrictive ventilatory disorders. We hypothesized that nocturnal nasal ventilation (NNV) would bring about similar improvements in patients with severe chronic obstructive pulmonary disease (COPD). The study used a randomized, crossover design, with subjects undergoing NNV or "standard care" for sequential 3-month periods. Of 23 patients with obstructive lung disease and a FEV1 less than 1 L who were initially enrolled, 4 were excluded because of obstructive sleep apnea prior to randomization. Among the remaining 19 patients, 7 withdrew because of intolerance of the nose mask, 5 were withdrawn because of intercurrent illnesses, and 7 completed both arms of the protocol. These latter 7 patients used the ventilator for an average of 6.7 h/night, and 3 of the 7 had partial relief of dyspnea during ventilator use. However, in comparison with studies performed upon initiation or after the standard care arm of the study, studies performed after 3 months of NNV revealed no improvements in pulmonary function, respiratory muscle strength, gas exchange, exercise endurance, sleep efficiency, quality or oxygenation, or dyspnea ratings. The only improvements observed were in neuropsychological function, possibly related to a placebo effect or another unknown mechanism. Despite the small sample size, our study indicates that NNV is not well tolerated by and brings about minimal improvements in stable outpatients with severe COPD.