We hypothesized that mortality is higher in patients with congestive heart failure (CHF) who develop Cheyne-Stokes respiration (CSR) during sleep than CHF patients without CSR. Overnight polysomnography was performed on 16 male patients with chronic, stable CHF: nine had CSR during sleep (CSR group) and seven did not (CHF group). The CSR group had a higher apnea-hypopnea index (AHI: 41 +/- 17 versus 6 +/- 5/hr) and experienced greater sleep disruption. There were no significant intergroup differences between age, weight, cardiac function, and pulmonary function. After the initial sleep study, all patients were maintained on standard medical therapy for CHF without supplemental oxygen or nasal continuous positive airway pressure. Over the next 3.1 to 4.5 yr there was a significant difference between the number of deaths in each group. Five patients died in the CSR group and two received a heart transplant, whereas only one patient died in the CHF group. Regression analysis revealed that mortality was positively correlated with CSR, AHI, arousal index, and the amount of stage 1, 2 non-REM sleep and was inversely related to the total sleep time. We conclude that mortality is higher in CHF patients who develop CSR during sleep than CHF patients without CSR. Although the development of CSR may simply reflect more severe cardiac impairment, we suggest that CSR itself accelerates the deterioration in cardiac function.