Prolonged ventilatory support by phrenic nerve stimulation (diaphragm pacing) has been developed into a useful therapeutic technique over the past two decades. The benefits of diaphragm pacing to patients with respiratory insufficiency resulting from central alveolar hypoventilation, and respiratory paralysis associated with quadriplegia, have led physicians to consider diaphragm pacing in patients with respiratory insufficiency who are dependent on ventilatory support. Some of these patients are initially seen in the intensive care unit because of respiratory failure or arrest. We discuss the evaluation and selection of candidates for diaphragm pacing. For over 200 years physicians have attempted to ventilate patients artificially by electrically stimulating the phrenic nerves [1]. In 1950 Sarnoff and colleagues [2,3] ventilated patients with acute respiratory insufficiency resulting from bulbar poliomyelitis by stimulating the phrenic nerves with electrodes connected to percutaneous wires. However, this technique could not provide extended support because of the risk of infection inherent with percutaneous wires. Prolonged ventilatory support by phrenic nerve stimulation (diaphragm pacing) was first accomplished successfully in 1966 by Judson and Glenn [4], who developed an implantable radiofrequency phrenic pacemaker activated and powered by an external transmitter.In the past 20 years about 600 patients have been treated with diaphragm pacing for chronic ventilatory insufficiency [5]. The technique requires intrathoracic implantation of electrodes on the phrenic nerve. Although the electrode can be implanted on the phrenic nerve through a cervical incision, this is not the preferred approach because the electrode will not stimulate branches of the phrenic nerve that join the main trunk in the thorax [5].The electrodes are connected to a subcutaneously implanted radiofrequency receiver. An external radiofrequency transmitter provides power to the receiver, generating a series of direct current pulses to the phrenic nerve. Respiratory rate and the amplitude of diaphragm contraction are controlled by adjusting the external transmitter (Figure). Refinements of the technique have decreased the incidence of phrenic nerve damage during operations, and have reduced the problem of diaphragm fatigue [5,6]. In many patients with respiratory paralysis from quadriplegia, bilateral diaphragm pacing [6] has eliminated or reduced the need for mechanical ventilation from a positivepressure ventilator, allowing these patients greater mobility and independence [6][7][8][9][10][11]. Unilateral diaphragm pacing has been used for patients with central alveolar hypoventilation (CAH) [10]; however,