Three sound recordings of tornadoes have been spectrally analyzed over the frequency interval between 100 and 2000 Hz. The low‐frequency analysis was limited by the response of the microphones used for the recordings; the upper frequency limit was imposed both the microphone response and the low‐signal level. One recording was made by James Cramer of a tornado which passed through Clay Center, Kansas on 25 September 1973. A second recording was made by Richard Allen Lindley of a tornado that passed through Guin, Alabama on 3 April 1974, and a third recording was made by Tom Bittman of a tornado which damaged his home in Tulsa, Oklahoma on 8 June 1974. All of the recordings are of low quality and the audio information required major spectral corrections, but the data analysis does indicate that the audio emissions from the tornadoes decrease in intensity as a function of increasing frequency. An attempt has been made to correlate the data with real (and conjectured) physical characteristics of the tornadoes and attendent atmospheric phenomena. It is clear to us from an objective analysis (and the more difficult subjective evaluation of the recordings) that identification of tornadoes based on acoustic emissions is possible . A study of tornadic sounds can not only provide a new tool for gaining insight into electrical and mechanical disturbances within a tornadic storm, but will also allow acoustic detection of a tornadic storm. A study of the change in intensity of the sounds emitted by the approaching Guin storm at both high and low frequencies suggests that noises generated by the high‐speed winds of the principal tornado vortex as its base scoured the ground might be discerned from noises generated by the winds of one or more smaller vortices moving around the tornado, and/or electrical discharges aloft.Subject Classification: [43]28.45, [43]28.65.