Auditory evoked responses were recorded in 16 normally hearing subjects in order to investigate the mechanisms underlying the generation of the 40 Hz steady-state response (SSR). In the first part of our study, auditory potentials were evoked by 0.1 ms clicks presented at 105 dB p.e. SPL with repetition rates of 7.9 (to obtain middle latency response, MLR), 20, 30, 40, 50, 60 Hz. In each subject predictions of the responses recorded at stimulus repetition rates of 30, 40, 50, 60 Hz were synthesized by superimposing MLRs at suitable time intervals. The calculated mean amplitude/rate and phase/rate functions behaved similarly for the recorded and predicted curves, showing the highest amplitude at 40 Hz and a linear increase of phase values when increasing the stimulus rate. Nevertheless the synthetic curves closely predicted amplitude and phase values of the recorded responses only at 40 Hz. At frequencies below 40 Hz, the mean amplitude of the predicted curve was lower than that of the recorded one while at frequencies above 40 Hz the mean amplitude was higher. Predicted phase values were found lagging at 30 Hz, and leading at 50 Hz and 60 Hz in comparison to phase values calculated on the recorded responses. Our findings suggest that a model based on the linear addition of transient MLRs is not able to adequately predict steady-state responses at stimulus rates other than at 40 Hz. Other mechanisms related to the recovery cycle of the activated system come into play in the steady-state response generation causing a decrease in amplitude and an increase in phase lag when increasing the stimulus repetition rate.