This study evaluated effects of auditory feedback upon production of helium speech. Recordings were made of five subjects who spoke on exhalations of air and an 80-20 mixture of He/O2, in quiet and with 95 dB SPL noise introduced to the ear via earphones. Talkers used a VU-meter to speak at a constant level throughout recording. The long-term average spectra of speech did not reveal striking differences between the two breathing mixtures as expected. Maximum energy was at 500 Hz for air and 900 Hz for He/O2. Spectrograms of vowels revealed typically reported shifts upward in the helium speech. Neither the average spectra nor the spectrograms of helium speech showed significant differences between talking in noise and quiet. Auditory feedback apparently is not utilized by talkers to change frequency characteristics of helium speech over short periods of time. On the other hand, when noise was present in air, several decibels more energy was observed at frequencies above 1300 Hz. Mean intelligibility scores significantly improved for both air and helium when noise interfered with a talker's ability to hear his speech. This improvement suggests that talkers try to compensate for their inability to hear themselves by speaking more distinctly.