Ten ear-protective devices—three plugs, three muffs, two phone-in-muff combinations, one phone-in-muff in protective helmet, and one plug-muff combination—were tested for noise attenuation. The psychophysical method of binaural threshold shift (occluded ears minus open ear) for half-octave bands of noise using three trials on five experienced listeners was used. When averaged over all frequencies at half-octaves (60–8000 cps), the V-51R plug under a Willson sound-barrier muff gave the best attenuation; the Sharpe HA-10 earphone, the Willson sound-barrier muff, and the David Clark 372-HPF muff gave almost equal and the next best attenuation; the Zwislocki resonant plug showed only slightly less attenuation than the three muffs; the V-51R and ComFit plugs were nearly equal with somewhat less attenuation than the Zwislocki resonant plug; the Gentex helmet and Telephonics sonibar muffs (with sound-powered phone and without) were the least effective attenuators of noise. Only in the frequency range from 60 to 300 cps were plugs better than the better muffs. All devices gave more attenuation as the frequency region increased from low (60 300 cps), to medium low (300 1000 cps), to medium high (1000–2900 cps), to high (2400–8000 cps), except for the Zwislocki resonant plug in the high-frequency region. The effect of sealing ring, headband pressure, and volume was experimentally studied independently of the actual devices tested. Flow-resistance measurements were made to explain the combined effects of headband pressure and seal around the ear.