The objective of this study was to evaluate the long-term cardiovascular mortality in patients with severe obstructive sleep apnoea (OSA) treated with either continuous positive airway pressure (CPAP) or mandibular advancing device (MAD). A non-concurrent cohort study of 570 subjects with severe OSA (apnoea/hypopnoea index (AHI) ≥ 30/h) and a control group of 269 subjects (AHI < 5/h) were followed up for a median of 79 months (interquartile range 76-88 months). All patients received CPAP initially. MAD was offered for those who were non-adherent to CPAP. The endpoint was cardiovascular death. Two hundred and eight control subjects, 177 patients treated with CPAP, 72 with MAD and 212 who declined treatment were analysed. Forty-two patients had a fatal cardiovascular event during the course of the study. The non-apnoeic group had the lowest cardiovascular death rate (0.28 per 100 person-years (95% confidence interval (CI): 0.08-0.71)) followed by the CPAP-treated (0.56 per 100 person-years (95% CI: 0.20-1.23)) and the MAD-treated OSA group (0.61 per 100 person-years (95% CI: 0.13-1.78)), with the highest cardiovascular mortality rate observed in the untreated OSA group (2.1 per 100 person-years (95% CI: 1.37-2.92)). Although residual AHI for MAD-treated patients was significantly higher than CPAP-treated patients (16.3 ± 5.1/h vs. 4.5 ± 2.3/h; P < 0.001), there was no difference in cardiovascular death rate between the two groups (hazard ratio 1.08 (95% CI: 0.55-1.74); P = 0.71). Both CPAP and MAD may be equally effective therapy in reducing the risk of fatal cardiovascular events in patients with severe OSA.