The intent of this study was to determine whether sound‐level measurements in dBAcould adequately depict harmfulness to hearing from exposures to noises differing widely in spectra. Each of 11 male listeners, ages 20–26 years, were exposed in separate sessions to each of three noises whose spectral slopes were −6, 0, and +6 dB per octave band and then re‐exposed to the same test noises in three additional retest sessions. All such noises were presented for 30 min via an earphone at 100 dBA. Analyses of variance of the temporary threshold shifts in hearing caused by these noise conditions [corrected to 2 min postexposure (TTS2)] showed insignificant differences, owing to the spectral variations for both test and retest conditions as averaged across audiometric frequencies 250–8000 Hz. Interactive effects between spectra and test frequencies were significant, the −6‐ and 0‐dB spectral conditions causing relatively more TTS2at frequencies below 3000 Hz, and the +6‐dB spectrum more TTS2at frequencies above 3000 Hz. Implications of these results for noise hazard rating schemes using dBAmeasures as contrasted with spectral determinations were noted.