Thoracic impedance apnea monitors may fail to detect obstructive apnea, may falsely alarm when the infant is breathing, and may confuse cardiac artifact with respiratory impedance. Therefore, we compared the performance of a respiratory inductive plethysmograph and a thoracic impedance monitor with a reliable measure of airflow, either nasal CO2 or pneumotachograph, during 29 studies in 28 patients referred for sleep laboratory evaluation. Sleep time averaged 72 +/- 37 (SD) minutes. The inductance plethysmography and the impedance monitor detected 99.6% +/- 0.6% and 98.3% +/- 3.0% of breaths, respectively. However, in two studies, the impedance monitor detected many extra breaths, once because of cardiac-induced impedance changes and once because of partial airway obstruction-induced impedance changes. In 11 studies, cardiac artifact was sometimes misinterpreted as a breath by the impedance monitor. The impedance monitor, but not the inductance plethysmograph, missed breaths following sighs in 16 of 29 studies. Both monitors detected all 60 episodes of central apnea. The inductance plethysmography detected 35 of 38 episodes of obstructive apnea, but the impedance monitor identified only two such events. Apnea was detected falsely four times by the inductance plethysmograph and 14 times by the impedance monitor. These results suggests that a respiratory inductive plethysmograph would have significant advantages over impedance monitoring, including the ability to detect obstructive apnea, and freedom from cardiac artifact.