An analytic compound action potential (CAP) obtained by convolving functional representations of the post-stimulus time histogram summed across auditory nerve neurons [P (t)] and a single neuron action potential [U (t)] was fit to human CAPs. The analytic CAP fit to pre- and postnoise-induced temporary hearing threshold shift (TTS) estimated in vivo P (t) and U (t) and the number of neurons contributing to the CAPs (N). The width of P (t) decreased with increasing signal level and was wider at the lowest signal level following noise exposure. P (t) latency decreased with increasing signal level and was shorter at all signal levels following noise exposure. The damping and oscillatory frequency of U (t) increased with signal level. For subjects with large amounts of TTS, U (t) had greater damping than before noise exposure particularly at low signal levels. Additionally, U (t) oscillation was lower in frequency at all click intensities following noise exposure. N increased with signal level and was smaller after noise exposure at the lowest signal level. Collectively these findings indicate that neurons contributing to the CAP during TTS are fewer in number, shorter in latency, and poorer in synchrony than before noise exposure. Moreover, estimates of single neuron action potentials may decay more rapidly and have a lower oscillatory frequency during TTS.