In the verbal domain, memory performance declines with consecutive tests of memory using similar stimuli. This decline in memory performance due to prior trials is called proactive interference (PI). Although PI is thought to reflect an important general limit of working memory function, little is known about PI in nonverbal domains. Using a short-term recognition task that reliably elicits PI in the verbal domain, we investigated the behavioral and neurophysiological effects of prior similar trials on musical working memory. It was predicted that memory performance would not decline with consecutive tests using similar trials, and may even improve as similar trials accumulated. Event-related potentials (ERPs) were expected to be inconsistent with interference from previous trials, and possibly reflect a decrease in memory load with similar prior trials. EEG was recorded while subjects performed a musical version of the Sternberg recognition memory task. Three chords were presented, followed by a short delay and a probe chord that matched one of the three chords of the set (inset) or none of them (outset). The chords were presented in blocks of familiarity (either familiar or unfamiliar to Western listeners), and subjects were grouped according to musical experience (musician, nonmusican). To test the effect of similar previous trials, a condition with three consecutive trials in the same key was contrasted with a condition where the key changed each trial. Auditory ERPs (N100 and P200), frontal slow waves, N400, and a late positive wave (LPW) were measured across the encoding, retention, and retrieval phases. Memory performance did not exhibit PI across the consecutive same-key trials. For the inset probes, accuracy was higher in the second and third trials within the three trials of the same key, showing memory facilitation. Prior similar trials had no effect on the auditory ERPs. Prior similar trials increased the amplitude of right frontal slow negative (SFN) waves during encoding and retention, and increased the amplitude of an SFN and a right-lateralized LPW during retrieval for the inset probes. During encoding the SFN was largest during the first trial of the three consecutive same-key trials. Musicians were more accurate, and responses were faster in the unfamiliar condition. Over the left hemisphere during encoding and retention frontal slow waves were more negative for musicians, especially in the familiar condition. During the retrieval phase musicians had equivalent amplitude LPWs for inset and outset probes, while nonmusicians had a larger LPW to the inset than outset probes. In conclusion, both behavior and event-related potentials were consistent with a lack of PI and with working memory facilitation due to prior similar trials. A right-lateralized network mediated the effect of prior similar trials in all phases of musical memory. Memory of previous similar trials influenced the top-down aspects of working memory and not perceptual processes, indicating STM-LTM interactions at a relatively late stage of processing. Event-related potentials suggest that one mechanism of musical working memory facilitation from prior trials may be enhanced retrieval of the set information