Birdsong consists of species-specific learned vocal sequences that are used primarily to attract mates and to repel competitors during the breeding season. However, many birds continue to sing at times when vocal production has no immediate or obvious impact on conspecific behavior. The mechanisms that ensure that animals produce important behaviors in contexts in which the function of these behaviors is not immediate or obvious are not known. One possibility is that animals engage in such behaviors because they are associated with pleasure. Here we examined the hypothesis that male European starlings sing outside of the breeding season in part because the act of singing in this context is facilitated and/or maintained by opioid-mediated reward. We measured song-associated reward using a conditioned place preference (CPP) test in male starlings producing fall, non-breeding season-typical song. We used quantitative real time PCR to measure expression of the enkephalin opioid precursor preproenkephalin (PENK) and mu opioid receptors (MOR) in the medial preoptic nucleus (POM; a region in which opioids are implicated in both reward and starling fall song) and additionally the song control region HVC as a control. Starlings developed a strong preference for a place that had been paired previously with the act of producing fall-typical song, indicating that fall song production was associated with a positive affective state. Both PENK and MOR mRNA expression in the POM, but not HVC, correlated positively with both individual reward state (as reflected in CPP) and undirected singing behavior. These results suggest that singing induces opioid receptor and enkephalin expression in the POM and consequent reward, and/or that opioid release in the POM induced by individual or environmental factors (e.g., the presence of food, safety of a flock or the absence of predators) induces a positive affective state which then facilitates singing behavior.