Previous research reported that people can successfully determine the winner of a musical competition when viewing a six second film clip of the performer without sound (Tsay, 2013, 2014); in contrast, when given an audio-only film clip or a clip that combined auditory and visual information, people perform at chance. Given the well-known publication bias in psychology (Ioannidis, 2005), this surprising and counterintuitive finding begs replication. In Study 1, 112 participants were randomly assigned to a sound, video, or video-plus-sound condition and were asked to select the winning musician after viewing five pairs of clips, one showing the winner and the other showing a non-winning musician. Clips were presented for 60 instead of six seconds, with the goal of giving participants more information about the performance, a modification we predicted would enhance performance in the audio and audio-visual conditions. Contrary to Tsay (2013), participants performed at chance in all three conditions. To more directly replicate Tsay (2013), in Study 2, 69 additional participants were randomly assigned to either a sound, video, or sound plus video condition and were asked to select the winning musician after viewing five pairs of 6-second clips showing the winner and another, non-winning musician. Here again the results did not replicate Tsay (2013): Participants performed significantly above chance in only one condition – when only hearing the performance and not seeing it. These results suggest that previous findings showing increased performance in rating musical performances without sound may be spurious and due to sampling error, issues in experimental design, low power, publication bias, or some combination of these. This also shows the strong importance of replication studies.