We studied 134 patients with Steinert's myotonic dystrophy (MD) in order to determine the prevalence of chronic hypercapnia, the level of muscle weakness and forced expiratory volume at which hypercapnic respiratory failure is likely to occur, and how clinical assessment might help predict hypercapnic respiratory failure. Subjects were divided into five classes with a muscular disability rating scale (MDRS): 0 = no clinical impairment (n = 9), I = minimal signs of impairment (n = 11), II = distal weakness (n = 41), III = moderate proximal weakness (n = 62), and IV = nonambulatory (n = 11). The prevalence of hypercapnia (PaCO2 > or = 43 mm Hg) was found to be 0%, 27%, 29%, 45% and 55% for MDRS 0 to 4, respectively (p = 0.03). A multiple regression analysis limited to clinical data showed that daytime hypersomnolence was a significant cofactor with the MDRS (p = 0.01) in predicting PaCO2 (r = 0.40). Among respiratory parameters, FVC, respiratory muscle strength (RMS), and maximal inspiratory pressure against occluded airways (PImax) were found to be predictors of nearly equal strength, explaining 16%, 15%, and 14% of the PaCO2 variance, respectively. In multiple regression analysis, sex, daytime sleepiness, and the expected/observed FVC ratio for a given RMS were found to be significant cofactors with PImax in predicting PaCO2 (r = 0.51). It is concluded that respiratory insufficiency should be suspected in MD patients with proximal weakness or daytime sleepiness. The likelihood of hypercapnia also increases with volume restriction and respiratory muscle weakness. Our study suggests that the combination of inspiratory muscle weakness and loading plays a predominant role in the pathogenesis of chronic alveolar hypoventilation in MD patients. The occurrence of daytime hypersomnolence suggests that other factors, such as low central ventilatory drive or sleep apnea, might play an additional role.