A miniature microphone was placed in a subject's ear after the method described by Brammer and Piercy [J. Acoust. Soc. Amer. 52, 1972 180], and its output fed to a spectrum analyzer set to a bandwidth of 100 Hz. The subjects were seated in a semireverberant room into which a white noise field at 95 dB SPL was introduced. The attenuation of three ear covering devices was measured by subtracting the spectrum obtained with the ear cover in place from the spectrum obtained with the subject's ears bare. The repeatability of measurements was examined with both small changes in microphone and ear cover placement. Microphone placement proved to be more critical, although correlation coefficients between measures across days averaged 0.88. Differences between runs measured at selected frequencies between 250 Hz and 7 kHz, generally fell within a range of ±3 dB. Validity was estimated by comparing the microphone data with psychophysical data gathered earlier. The former tended to indicate more attenuation than the latter, though the correspondence between the two was encouraging.