We studied the relationship between auditory activity in the midbrain and selective phonotaxis in females of the treefrog, Pseudacris crucifer. Gravid females were tested in two-stimulus playback tests using synthetic advertisement calls of different frequencies (2600 versus 2875 Hz; 2800 versus 3500 Hz; 2600 versus 3500 Hz). Tests were conducted with and without a background of synthesized noise, which was filtered to resemble the spectrum of a chorus of spring peepers. There were no significant preferences for calls of any frequency in the absence of background noise. With background noise, females preferred calls of 3500 Hz to those of 2600 Hz. Multi-unit recordings of neural responses to synthetic sounds were made from the torus semicircularis of the same females following the tests of phonotaxis. We measured auditory threshold at 25 frequencies (1800-4200 Hz) as well as the magnitude of the neural response when stimulus amplitude was held constant and frequency was varied. This procedure yielded isointensity response contours, which we obtained at six amplitudes in the absence of noise and at the stimulus amplitude used during the phonotaxis tests with background noise. Individual differences in audiograms and isointensity responses were poorly correlated with behavioural data except for the test of 2600 Hz versus 3500 Hz calls in noise. The shape of the neural response contours changed with stimulus amplitude and in the presence of the simulated frog chorus. At 85 dB sound pressure level (SPL), the level at which females were tested, the contours of females were quite flat. The contours were more peaked at lower SPLs as well as during the broadcast of chorus noise and white noise at an equivalent spectrum level (45-46 dB/Hz). Peaks in the isointensity response plots of most females occurred at stimulus frequencies ranging from 3200 to 3400 Hz, frequencies close to the median best excitatory frequency (BEF) of 3357 Hz but higher than the mean of the mid-frequency of the male advertisement call (3011 Hz). Addition of background noise may cause a shift in the neural response-intensity level functions. Our results highlight the well-known nonlinearity of the auditory system and the danger inherent in focusing solely on threshold measures of auditory sensitivity when studying the proximate basis of female choice. The results also show an unexpected effect of the natural and noisy acoustic environment on behaviour and responses of the auditory system. Copyright 1998 The Association for the Study of Animal Behaviour.