BACKGROUND Respiratory instability in chronic heart failure (CHF) is characterized by irregularly rapid respiration or non-periodic breathing rather than by Cheyne-Stokes respiration. We developed a new quantitative measure of respiratory instability (RSI) and examined its independent prognostic impact upon CHF. METHODS In 87 patients with stable CHF, respiratory flow and muscle sympathetic nerve activity (MSNA) were simultaneously recorded. RSI was calculated from the frequency distribution of respiratory spectral components and very low frequency components. RESULTS During a mean follow-up of 85±38 months, 24 patients died. Sixteen patients who died of cardiac causes had a lower RSI (16±6 vs. 30±21, p<0.01), a lower specific activity scale (4.3±1.4 Mets vs. 5.7±1.4 Mets, p<0.005), a higher MSNA burst area (16±5% vs. 11±4%, p<0.001), and a higher brain natriuretic peptide (BNP) level (514±559pg/ml vs. 234±311pg/ml, p<0.05) than 71 patients who did not die of cardiac causes. Multivariate analysis revealed that RSI (p=0.015), followed by MSNA burst area (p=0.033), was an independent predictor of subsequent all-cause deaths and that RSI (p=0.026), MSNA burst area (p=0.001), and BNP (p=0.048) were independent predictors of cardiac deaths. Patients at very high risk of fatal outcome could be identified by an RSI<20. CONCLUSIONS The daytime respiratory instability quantified by a new measure of RSI has prognostic importance independent of sympathetic nerve activation in patients with clinically stable CHF. An RSI of <20 identifies patients at very high risk for subsequent all-cause and cardiovascular death.