The purpose of this study was to test the ability of cardiopulmonary exercise testing (CPET)-derived variables as sudden cardiac death (SCD) predictors.The CPET variables, such as peak oxygen uptake (VO2), ventilatory requirement to carbon dioxide (CO2) production (VE/VCO2) slope, and exercise oscillatory breathing (EOB), are strong predictors of overall mortality in chronic heart failure (CHF) patients. Even though up to 50% of CHF patients die from SCD, it is unknown whether any of these variables predicts SCD.One hundred fifty-six CHF patients (mean age: 60.9 +/- 9.4 years; mean ejection fraction: 34.9 +/- 10.6%) underwent CPET. Subjects were tracked for sudden versus pump-failure cardiac mortality over 27.8 +/- 25.2 months.Seventeen patients died from SCD, and 17 died from cardiac pump failure. Survivors showed significantly higher peak VO2 (16.8 +/- 4.5 ml x kg(-1) x min(-1)) and lower VE/VCO2 slope (32.8 +/- 6.4) and prevalence of EOB (20.3%), compared with subjects who experienced arrhythmic (13.5 +/- 3.2 ml x kg(-1) x min(-1); 41.5 +/- 11.4; 100%) or nonarrhythmic (14.1 +/- 4.7 ml x kg(-1) x min(-1); 38.1 +/- 7.3; 47.1%) deaths (p < 0.05). At Cox regression analysis, all variables were significant univariate predictors of both sudden and pump failure death (p < 0.01). Multivariate analysis, including left ventricular (LV) ejection fraction, LV end systolic volume, and LV mass selected EOB, was the strongest predictor of both overall mortality (chi-square: 38.7, p < 0.001) and SCD (chi-square: 44.7, p < 0.001), whereas VE/VCO2 slope was the strongest ventilatory predictor of pump failure death (chi-square: 11.8, p = 0.001).Exercise oscillatory breathing is an independent predictor of SCD in patients with CHF and might help as an additional marker for prioritization of antiarrhythmic strategies.