The relationship between the calibration tone and speech volume unit (VU) levels for nine commercially recorded speech tests was studied. Two procedures were used. In procedure 1, experienced clinicians reported observations of four different audiometer VU meters responding to six commercially available speech tests. Differences in magnitude estimates among judges were usually about 1 dB or less. Peak VU level deviations from the zeroed calibration tone were within k1.5 VU except for the Fulton Lewis Jr. recording, which varied by about −4.8 VU. In procedure 2, a video system was used to record precisely VU meter responses to clinical speech materials. Differences between the calibration tones and speech levels were similar to those obtained using multiple observers. Statistically significant (p 5 0.01) differences were found between the speech levels of the commercial test recordings.