This study evaluated the characteristics of transient evoked otoacoustic emission (TEOAE) time-frequency (TF) representations from normal and hearing-impaired ears. Linear and non-linear TEOAE recordings from normally-hearing subjects (40) and non-linear recordings from patients with sensorineural hearing loss (SNHL) (40) were analysed using the short-time-Fourier-transform spectrogram, the Gabor spectrogram, and the adaptive spectrogram. The TF representations of the TEOAE recordings indicated a considerable dispersion of energy across frequencies and TEOAE time segments >4.0 ms. The linear and non-linear recordings from the normal subjects showed common frequency peaks. The TF representations from the patients with SNHL indicated that the significantly reduced energy in the mid-to-high TEOAE frequencies did not correlate closely with the threshold elevation. As in the recordings from the normal subjects, a high percentage of the TEOAE cumulative energy was found within a short TEOAE segment (4-14 ms).