Observations of the suppression of stimulus frequency evoked otoacoustic emissions (SFOAEs) by a second tone were made in human subjects. Measurements were made with the suppressor tone at frequencies and levels at, above, and below the stimulus tone generating the SFOAE. Data were collected with the stimulus tone at a range of levels (20-60 dB SPL), over a range of frequencies and in six different ears, in three subjects. The results were fitted to a phenomenological model, which allowed the data to be expressed as two parameters against the frequency of the suppressor tone. Characteristics of the suppression were examined, in particular the asymmetry between suppression by tones of higher and lower frequency than the tone evoking the SFOAE. At relatively low levels of suppressor tone, suppressors with frequencies higher than the stimulus tone were more effective suppressors than lower frequencies. At higher levels of suppressor tone, the situation was reversed, with lower frequencies being more effective than higher frequencies. These results were discussed in terms of nonlinear interaction between waves in the cochlea. This interpretation was used to estimate the shape of the traveling wave envelope produced by the stimulus tone, from the results of the suppression experiments. It was shown that the estimates of the shapes of the traveling wave envelope were nonlinear, the peak of the envelope becoming sharper at lower levels of stimulus. A simple quantitative model of SFOAE suppression was formulated using concepts of energy flow within the cochlea. This model produced SFOAE suppression results with all the major characteristics of SFOAE suppression from a real, human ear.