The effect of emotion on memory is powerful and complex. While there seems to be agreement that emotional arousal generally increases the likelihood that events are remembered, it is somewhat disputed whether also the valence of emotions influences memory. Specifically, several experiments by Kensinger and colleagues have provided evidence for the hypotheses that negative valanced emotions facilitate the encoding of perceptual details. On the other hand, Mather and colleagues have suggested that these results could be explained by confounding relationships of valence and arousal, i.e., that items that generate negative emotions are typically also more arousing. In this study, we provide a conceptual replication of Kensinger's findings. We employed a novel experimental design, in which the effects of standardized emotional arousing sounds on recognition accuracy for neutral visual scenes was measured. We indirectly manipulated the amount of visual detail that was encoded, by requiring participants to memorize either single exemplars (low interference) or multiple exemplars (high interference) of visual scene categories. With increasing visual overlap in the high interference condition, participants were required to encode a high degree of visual detail to successfully remember the exemplars. The results obtained from 60 healthy human participants confirmed Kensinger's hypothesis by showing that under conditions of high visual interference, negative valanced emotions led to higher levels of recognition accuracy compared to neutral and positive emotions. Furthermore, based on the normative arousal ratings of the stimulus set, our results suggest that the differential recognition effect cannot be explained by differing levels of arousal.