Sleep hypoventilation is an inevitable consequence of Duchenne muscular dystrophy (DMD), usually preceding daytime respiratory failure. Appropriate scheduling of polysomnography and the introduction of noninvasive ventilation (NIV) during sleep are not defined. Our aim was to determine the parameters of daytime lung function associated with sleep hypoventilation in patients with DMD. As our method we chose a prospective comparison of wakeful respiratory function (spirometry, lung volumes, maximal mouth pressures, arterial blood gases) with outcomes of polysomnography. All measurements were made with subjects breathing air. Nineteen subjects were studied. The FEV(1) was correlated with Pa(CO(2)) (r = -0.70, p < 0.001) and base excess (r = -0.68, p < 0.01). All of these parameters were significantly related to sleep oxygenation (proportion of total sleep time spent at an Sa(O(2)) </= 90% [TST < 90%]). An FEV(1) < 40% was a sensitive (91%) but not specific (50%) indicator of sleep hypoventilation (TST < 90% of >/= 2%); a Pa(CO(2)) of >/= 45 mm Hg was an equally sensitive (91%) but more specific (75%) indicator while a base excess of >/= 4 mmol/L was highly specific (100%) but less sensitive (55%). After introduction of NIV during sleep (n = 8), there was a significant reduction in wakeful Pa(CO(2)) (54 +/- 7.4 to 49.1 +/- 4 mm Hg, p < 0.02) over 0. 9 +/- 0.4 yr despite a further decline in FEV(1) (0.84 +/- 0.46 to 0. 64 +/- 0.39 L, p < 0.05). We conclude that in patients with DMD, (1) arterial blood gases should be performed once the FEV(1) falls below 40% of the predicted value; (2) polysomnography should be considered when the Pa(CO(2)) is >/= 45 mm Hg, particularly if the base excess is >/= 4 mmol/L; (3) the decrease in wakeful Pa(CO(2)) after NIV administered during sleep implicates sleep hypoventilation in the pathogenesis of respiratory failure; and (4) impaired ventilatory drive is a possible mechanism for respiratory failure, as the NIV-associated decrease in wakeful Pa(CO(2)) occurs despite a further decline in ventilatory capacity, suggesting continuing deterioration in respiratory muscle function.