Chronic heart failure (CHF) patients frequently show sleep-disordered breathing consisting of periodic breathing (PB) and Cheyne-Stokes respiration (CSR) with central sleep apnoea (CSA). Since the diagnosis of sleep-disordered breathing, in CHF patients, can be made only by means of full polysomnography, the aim of the present study was to evaluate whether or not daytime respiratory function can identify patients at risk of nocturnal PB and/or CSR/CSA. Twenty-seven patients (mean age 54 +/- 8.5 yrs), eight New York Heart Association Functional Class (NYHAFC) II, 17 NYHAFC III and two NYHAFC IV, with severe cardiac failure (cardiac output 2.0 +/- 0.66 L.min-1, ejection fraction 22.5 +/- 5.77%, pulmonary capillary wedge/pressure 23 +/- 9.05 mmHg). Mouth occlusion pressure (P0.1)/maximal inspiratory pressure (MIP) was significantly higher in patients with nocturnal CSR/CSA (5.04 +/- 1.49 versus 3.24 +/- 2.13%, analysis of variance (ANOVA) 0.03), whereas their arterial carbon dioxide tension (Pa,CO2) was significantly lower (4.15 +/- 0.56 (31.2 +/- 4.23 mmHg) versus 4.67 +/- 0.53 kPa (35.1 +/- 4 mmHg), ANOVA 0.02). Logistic regression analysis demonstrated that CSR/CSA occurrence may be predicted by daytime measurement of P0.1/MIP and Pa,CO2 (p = 0.04 and 0.01 respectively; odds ratio 1.93 and 0.76 respectively). The sensitivity was 70%, specificity 76.5%, false positive rate 36.4%, false negative rate 18.8%, positive predictive value 71.4% and negative predictive value 85%. This model seems useful for predicting respiratory pattern changes in chronic heart failure patients and the authors suggest that polysomnography be performed only in high-risk patients, saving costs and the resources of sleep laboratories.