We have previously observed large phasic variations of respiratory mechanical impedance in chronic obstructive pulmonary disease (COPD) patients mechanically-ventilated for acute respiratory failure, and postulated that they were due to expiratory now limitation (EFL). The aim of this study was to test that assumption experimentally and to assess the value of impedance for automatic and noninvasive detection of EFL during mechanical ventilation. The study was performed: 1) in a mechanical analogue, including a flow-limiting element; and 2) in eight anaesthetized and paralysed rabbits, before and during histamine infusion, In both instances, En was obtained by lowering the expiratory pressure, using a computer-controlled ventilator; the absence of flow increase when expiratory pressure was further lowered was taken as evidence of En, Impedance was measured by applying 15 Ri oscillations at the airway opening, Its real (Re) and imaginary (Im) parts were measured separately during the inspiratory and the expiratory phases, and their differences were related to the mean inspiratory modulus. With the analogue, EFL was accompanied by large decreases both of Re and Im during the expiratory phase, In the rabbits, phasic variations of Re were variable in sign and were not significantly different with and without En, In contrast, EFL systematically and specifically decreased Im during the expiratory phase, A threshold of -50% provided a sensitivity of 96% and a specificity of 100% for detecting En, The observed phasic variations may be explained by ah-way wall shunt properties. The study suggests that a large decrease of the imaginary part of impedance during the expiratory phase is a sensitive and specific index of expiratory now limitation during artificial ventilation.