It would be reasonable to expect that our previous experience regarding a stimulus that predicts harm would make the subsequent identification of that stimulus easier when harm happens again. Forty-eight volunteers were submitted to both phases of this sequence of events: learning of the predictive relationship and later priming. A face with neutral expression (CS+) was paired with a moderately aversive electric shock and another (CS−) with a neutral tone. Subsequently, these two faces, as well as other known and new faces, were presented for familiarity judgments. Both the CS+ and the CS− faces were preceded by an aversive stimulus (aversive prime) in one occasion and by a neutral stimulus (neutral prime) in another. The familiarity judgment regarding the CS+ was faster after the aversive prime than after the neutral prime, but there was no difference regarding the CS−. The differential effect of the aversive prime over the CS+ and the CS− showed a significant but small correlation with the differential skin conductance response to CS+ and CS− (signal learning), and with the differential evaluation of those stimuli in terms of like-dislike (evaluative learning). The scope of these results, as well as the usefulness of this methodological model, is discussed.