The influence of one pure tone on the threshold of another was investigated. In contrast to previous experiments, masking in the present experiment was studied by determining the level of the masker necessary to mask a signal as a function of the frequency of the masker. The level and frequency of the signal served as parameters. The general forms of the masking functions are similar to those reported previously, but vary in several details. The maximum masking effect occurred when the frequency of the masker approximated that of the signal. When the masker frequency was greater than that of the signal the slope of the masking function was very steep, 150–280 db/octave depending upon signal level and frequency. This slope tended to be steeper at higher signal levels and, although less marked, also at higher signal frequencies. When the masker frequency was about 0.85 that of the signal an irregularity appeared in the masking curves. The size of this hump increased as signal frequency and level increased. Its frequency location seemed to rule out aural harmonics as a cause and its presence was tentatively related to perception of envelope modulation. The data are discussed from the points of view of band-pass filter analog and compared to studies of previous investigators. In general the present findings represent an extension of pure tone masking data with wider ranges of stimuli and a different method than those previously used.