Recently we proposed a theoretical model for cochlear nonlinearity and second filter [J. Acoust. Soc. Am.59, 408–423 (1971)]. From psychophysical two‐tone suppression data we attempted to estimate characteristics of the two filters and of the compressive nonlinearityy = xν. In our first experiments probe frequency (suppressee as well as test probe) was 1 kHz. Masker (suppressor) frequency was a parameter. We used the pulsation‐threshold method, where masker plus suppressee and test probe alternated. The pulsation threshold of the test probe was measured as a function of masker level, suppressee level being a fixed parameter. Data curves typically are characterized by a horizontal part, a descending part, and an ascending part. The descending part, reflecting suppression contains the wanted information. Insofar as our interpretation of the data is valid, we estimate for ν a value from 0.25 to 0.8, depending on musket frequency. The low‐frequency part of the second‐filter characteristic levels off at approximately 10‐dB attenuation. The first‐filter characteristic depends on level. Thus, the data show more nonlinearity than the model accounts for. Possible modifications will be discussed.