Trace eyeblink classical conditioning was assessed in patients with bilateral medial-temporal amnesia and matched control participants who had previously shown equivalent delay eyeblink conditioning (J. D. E. Gabrieli et al., 1995). The silent trace interval varied for durations of 500, 750, or 1,000 ms in successive sessions separated by at least 2 weeks; extinction trials followed each session. Patients with amnesia produced significantly fewer conditioned responses (CRs) than did control participants at all trace intervals. Both groups produced fewer CRs as the trace interval lengthened. Thus, the temporal lobe memory system in humans makes an essential contribution to normal acquisition in trace, but not delay, classical eyeblink conditioning.