Heart failure (HF) is a leading cause of mortality in Western societies.1– It is now recognized that Cheyne-Stokes respiration (CSR) during sleep, characterized by a cyclic pattern of waxing and waning of ventilation interposed by central central apneas or hypopneas, is very common among patients with severe HF.2–3 CSR is associated with oscillations in BP, heart rate, recurrent hypoxemia, arousals from sleep, increased sympathetic activity, ventricular arrhythmias, and increased risk of death.3–5 CSR is an exaggerated form of periodic breathing (PB). PB is also common in HF patients and is independently associated with poor prognosis when present during exercise.6 CSR is a manifestation of respiratory instability and is particularly prone to occur during sleep when the respiratory system becomes critically dependent on the metabolic control system.5 Numerous mathematical modeling studies5,7 have been proposed to explain CSR. The models include three basic components: controlling system, controlled system, and a feedback loop. The controlled variables are Pao2 and Paco2. The stability of the system may be tested by subjecting it to small changes in ventilation (termed disturbance), to which the controller responds with a corrective action in ventilation (termed correction). The correction/disturbance ratio is termed loop gain. Loop gains < 1 mean that the disturbance will be damped, whereas loop gains > 1 will result in PB due to amplification of the initial disturbance. Loop gain is dependent on three factors: the time lag between the disturbance and the response, controller gain, and plant gain. Prolonged circulation time, a hallmark of HF, results in a delayed ventilatory response that may, instead of damping the primary disturbance, promote respiratory instability. Because several patients with severe HF have presented stable breathing, there is a general agreement that circulatory delay contributes to but is not sufficient to cause CSR.,5,7 Controller gain is represented by chemoreceptor sensitivity. High sensitivity will promote exaggerated response, ventilatory overshoot, and blood gas instability. Increased controller gain is well documented and thought to play a central role in the genesis of CSR in patients with HF.5 The controlled system may act both as damping or exaggerating the ventilatory disturbances; this is termed plant gain. Plant gain is dependent on lung gas stores, body stores of O2 and CO2, and metabolic rate. For instance, reductions in lung volume increase plant gain because smaller lung volumes are less effective at damping out changes in Paco2 and Pao2, thus favoring instability.,7– The effects of plant gain on the genesis of CSR have been alluded to in theoretical studies but have received little attention in experimental studies. In this issue of CHEST (see page 67), Szollosi et al8 evaluated 45 consecutive HF patients referred for polysomnography who underwent spirometry and measurement of arterial blood gas (ABG) levels (while awake) and single-breath diffusion capacity for carbon monoxide (Dlco). After excluding patients with stable breathing and patients with predominantly obstructive sleep apnea, apnea-hypopnea index (AHI) was correlated with spirometry, Dlco, and ABG in the remaining 22 patients. AHI correlated with Dlco, Pao2, and Paco2. In a forward stepwise linear regression model, Dlco and Pao2, but not Paco2, remained significant and accounted for 49% of the supine AHI variability. The authors concluded that impaired gas exchange (representing an increased plant gain) may contribute to CSR in patients with HF.,8 This theory raises the importance of hypoxia in the genesis of CSR. In contrast to O2, CO2 is a more diffusible gas and will be less affected by changes is Dlco. Moreover, body stores of O2 are lesser than those of CO2, and therefore oscillations in Pao2 are more prone to occur.,5,7 Following this line of thinking, oscillation in Pao2 will be amplified by a low Dlco and will in turn contribute to respiratory instability in patients with HF.