Aural harmonic distortion is generated in normal ears with a constant 1000‐Hz fundamental (f1) while the listener traces his threshold for an interrupted 2000‐Hz (f2) objective tone. Both tones are presented simultaneously to the same ear with the consequent variations in the masked thresholds recorded during manipulations of the time relation between f1 and f2. The technique and preliminary results are described in the first part of this report. These data show that the threshold shift (TS), produced with an f1 of 60 or 65 dB sensation level (SL), depends critically upon the phase relation of f2 to f1: The TS varies sinusoidally as f2 is phase shifted through one or two complete cycles. Section II presents a discussion of the assumptions used to explain these results. Accordingly, a harmonic series is generated within the ear, vectorial summation occurs between the aural harmonic (AH) and the objective tone, and the task of the listener is to maintain this resultant at some level. An electronic analog model shows that, under these assumptions, the objective f2 would undergo variations approximating sine waves if the AH is at a low effective level compared to that needed to attain audibility. The experiment of Section III confirmed that the general equation, Y = A1 + A2 sin (x+ϕ), provides a satisfactory description of the variations in TS. In addition, a prediction was derived concerning the phase conditions causing maximum interference, both constructive and destructive. The changes in TS associated with these two phase conditions are approximately equal and opposite as predicted. In fact, the objective f2 is heard at lower intensities than under quiet listening conditions when the f1 intensity was increased up to 60 dB (SPL). This rather bizarre reversal of the usual masking phenomenon, owing to augmentation between the AH and objective f2, was anticipated from the model.