Respiratory inductance plethysmography (RIP) has been widely used to measure ventilation during sleep, but its accuracy in this role has not been adequately tested. We have thus examined the accuracy of the RIP by comparing tidal volume measured with RIP with that measured by a pneumotachograph in eight unrestrained normal subjects during sleep. We have also studied the effect of posture on the accuracy of the RIP. In all sleep stages the correlation between RIP tidal volume measurements and expired volume showed relatively poor correlations (mean r = 0.49–0.60), and the bias of the measurements varied widely. Changes in posture altered the correlations between the two measurements, with no systematic differences between positions. When the subjects resumed a position, the 95% confidence intervals of tidal volume measurement did not overlap the original confidence limits in that posture on 13 of 25 occasions. This study shows that the RIP does not accurately measure tidal volume during sleep in unrestrained subjects and should only be used for semiquantitative assessment of ventilation during sleep.