The accuracy of tympanometric estimates of ear canal volume was evaluated by testing the following two assumptions on which the procedure is based: (a) ear canal volume does not change when ear canal pressure is varied, and (b) an ear canal pressure of 200 daPa drives the impedance of the middle ear transmission system to infinity so the immittance measured at 200 daPa can be attributed to the ear canal volume alone. The first assumption was tested by measuring the changes in ear canal volune in eight normal subjects for ear canal pressures between ±400 daPa using a manometric procedure based on Boyle's gas law. The data did not support the first assumption. Ear canal volume changed by a mean of .113 ml over the ±400 daPa pressure range with slightly larger volume changes occurring for negative ear canal pressures than for positive ear canal pressures. Most of the volume change was attributed to movement of the probe and to movement of the cartilaginous walls of the ear canal. The second assumption was tested by comparing estimates of ear canal volume from susceptance tympanograms with a direct measurement of ear canal volume adjusted for changes in volume due to changes in ear canal pressure between +±400 daPa. These data failed to support the second assumption. All tympanometric estimates of ear canal volume were larger than the measured volumes. The largest error (39%) occurred for an ear canal pressure of 200 daPa at 220 Hz, whereas the smallest error (10%) occurred for an ear canal pressure of ±400 daPa at 660 Hz. This latter susceptance value (-400 daPa at 660 Hz) divided lay three is suggested to correct the 220-Hz tympanogram to the plane of the tympanic membrane. Finally, the effects of errors in estimating ear canal volume on static immittance and on tympanometry are discussed.