By means of the articulation recordings and high‐fidelity reproducing system described in an earlier paper, and a large variety of frequency distortion networks, the selective amplification principle in hearing aids has been tested by extensive measurements on normal and typical hard‐of‐hearing persons. The criterion used for evaluating the effectiveness of each type of amplification has been the standard percent syllable articulation. For all ears tested, uniform amplification, free from nonlinear distortion and adventitious noise, has proved superior to amplification “peaked” at various frequencies to provide what is often falsely called selective amplification. Carefully prescribed selective amplification, even when first tried, has proved to be better than uniform in some instances. In other cases it does not give as good results. Some hard‐of‐hearing persons did not hear well witheitheruniform or selective amplification at first, because they never had heard some of the speech sounds before, or had become unfamiliar with them owing to their impairments. However, after practice with the prescribed amplification most of them became accustomed to it, and their percentage articulation rose with successive tests. The opinion of the listener has proved to be a very poor criterion for determining the correct amplification. Many of the hard‐of‐hearing listeners have chosen as sounding best the types which had frequency response curves of the same general shape as their threshold curves, and have criticized as sounding very unpleasant those with which they attained the highest articulations. It is not yet always possible to prescribe immediately the best type of amplification for a given hearing impairment, but a method of prescription based on the “most comfortable equal loudness curve” has proved highly satisfactory in several cases.