This study explored the time course of conditioned fear response expression. Two neutral male facial expressions served as conditioned stimuli (CS) in a differential trace conditioning that involved either an aversive (n=14) or a nonaversive (n=12) unconditioned stimulus (UCS) in a between-subjects design. Skin conductance response (SCR) to the CSs and startle response magnitudes to acoustic probes presented at early (250 ms) or late (1,750 ms) probe times after CS onset were measured. As expected, conditioned SCR discrimination was observed in both aversive and nonaversive learning, whereas the conditioned potentiation of the startle response was only observed for the aversive UCS condition. Interestingly, conditioned startle discrimination was specific for the later probe time. In contrast, conditioned fear potentiation of the startle response at the early probe time was equally pronounced for CS+ and CS-. These findings suggest that fear-eliciting neural structures are rapidly activated in fear learning, whereas the expression of inhibitory conditioning requires more time, presumably reflecting the involvement of cortical top-down control processes.