We measured the forward‐masked threshold for a 20‐ms, 1000‐Hz sinusoidal probe as a function of the levels of a 290‐ms, 1000‐Hz sinusoidal masker (30–70 dB SPL) and of a 539‐Hz sinusoidal suppressor 20–90 dB SPL) presented simultaneously with the masker. At the highest suppressor and musket levels, addition of the suppressor to the masker resulted in a 10‐dB reduction in threshold for the probe. These measurements were repeated with a broadband noise (low‐pass filtered at 5000 Hz, 20‐dB No) presented either during the masker interval or during both the masker and the probe intervals. When added during only the masker interval, the noise had no effect on the probe threshold except in conditions where the noise itself produced more forward masking than either sinusoid. When the noise was presented during both the masker and probe intervals, probe thresholds increased over the level predicted by a simple addition of masker effects. In these “continuous‐noise” conditions, adding the suppressor to the masker caused a reduction in probe threshold that was only sightly less than observed in the absence of the noise. We can account for these results by assuming that the nonlinear interaction of the forward masker and the noise occurs after the suppressor has reduced the effective level of the forward masker. [Research supported by NIH and NSF.]