Background and objective: Sleep disordered breathing (SDB) is highly prevalent but under-recognized. Evidence is accumulating for its role as a predictor of mortality independent of cardiovascular risk factors. The role of hypercapnia in SDB is not known systematically, and between 11 and 43% of patients with SDB develop chronic hypercapnia. Hypercapnia predicts mortality in other respiratory conditions. The role of hypercapnia in independently predicting mortality in patients with SDB after assessing for the presence of airways disease and obesity was investigated. Methods: The records of 396 consecutive patients were examined retrospectively. Univariate and multivariate analyses were performed using Cox proportional hazards regression to determine the association between gas exchange and polysomnography (PSG) variables and all-cause mortality, adjusted for potential confounders, including age, gender and presence of co-morbidity and airways disease. Results: The mean age of our patient population was 55 ± 15 years, and the mean body mass index (BMI) was 30.6 ± 6.2 kg/m 2. Out of the 322 patient population, 258 were men. After 10 years, 25% had died, following a maximum follow-up of 16.7 years. Mortality among patients with SDB was predicted by a difference between evening and morning PaCO 2 (ΔPaCO 2) of ≥7 mm Hg and evening hypoxaemia (PaO 2 < 65 mm Hg), independent of age and the presence of co-morbidity and airways disease. There was a significant correlation between minimum oxygen saturation (SpO 2) and ΔPaCO 2 ≥ 7 mm Hg (P = 0.002) and evening PaO 2 < 65 mm Hg (P < 0.001). Conclusions: An overnight increase in CO 2 and evening hypoxaemia are independent mortality predictors in SDB. A low minimum SpO 2 identifies patients in whom morning and evening arterial blood gases are beneficial.