Here, we examined how repetition under midazolam, a benzodiazepine that prevents the storage of novel associations, affects cued-recall performance of paired-associates. We contrasted word pairs that were initially studied and tested repeatedly without any successful recall prior to the midazolam injection, with other pairs that were studied for the first time after the injection of midazolam. According to our SAC (Source of Activation Confusion) memory model, repetition leads to strengthening existing memory traces rather than creating multiple traces for each repetition. As such, it predicts that repetition under midazolam should benefit only pairs that were originally studied prior to the midazolam injection. This prediction was confirmed. The results suggest that memory traces for pairs studied prior to the midazolam injection were strengthened under midazolam. However, word pairs that had not been studied prior to the injection were not bound in long-term memory because midazolam prevents the formation of new associations.