Chronic heart failure (CHF) is a major health problem associated with increased mortality, despite modern treatment options. Central sleep apnea (CSA)/Cheyne-Stokes breathing (CSB) is a common and yet largely under-diagnosed co-morbidity, adding significantly to the poor prognosis in CHF because of a number of acute and chronic effects, including intermittent hypoxia, sympathetic overactivation, disturbed sleep architecture and impaired physical tolerance. It is characterized by repetitive periods of crescendo-decrescendo ventilatory pattern, alternating with central apneas and hypopneas. The pathogenesis of CSA/CSB is based on the concept of loop gain, comprising three major components: controller gain, plant gain and feedback gain. Laboratory polysomnography, being the golden standard for diagnosing sleep-disordered breathing (SDB) at present, is a costly and highly specialized procedure unable to meet the vast diagnostic demand. Unlike obstructive sleep apnea, CSA/CSB has a low clinical profile. Therefore, a reliable predictive system is needed for identifying CHF patients who are most likely to suffer from CSA/CSB, optimizing polysomnography use. The candidate predictors should be standardized, easily accessible and low-priced in order to be applied in daily medical routine. The present review focuses on a pathophysiological approach to the selection of some predictors based on parameters reflecting the etiology, the pathogenesis and the consequences of CSA/CSB in CHF.