We examined the effects of processing fluency on metamemory for written music. In Experiment 1, piano players studied short sequences notated in either treble or bass clef by playing them on a silent keyboard with either their left or right hand, creating a congruent (fluent) and an incongruent (dysfluent) condition (hand/clef match or mismatch, respectively). A subsequent recognition test accompanied by confidence ratings (CRs) gauged retrospective metamemory. Items in the congruent conditions were recognized better (a desirable-difficulty effect), but CRs showed that participants were unaware of this memory difference. In Experiment 2, judgments of learning (JOLs) followed each studied sequence to gauge prospective metamemory. JOLs were higher in the congruent condition, although recognition was unaffected. In Experiment 3, whether the music was fingered on the silent keyboard or not did not influence results. These data are discussed within the framework of metacognitive theories that emphasize the importance of processing fluency.