It has long been noted that expert musicians lengthen notes at phrase boundaries in expressive performance. Recently, we have extended research on this phenomenon by showing that undergraduates with no formal musical training and children as young as 3 years lengthen phrase boundaries during self-paced listening to chord sequences in a lab setting (the musical dwell-time effect). However, the origin of the musical dwell-time effect is still unknown. Recent work has demonstrated that musicians and non-musicians are sensitive to entropy in musical sequences, experiencing high-entropy contexts as more uncertain than low-entropy contexts. Because phrase boundaries tend to afford high-entropy continuations, thus generating uncertain expectations in the listener, one possibility is that boundary perception is directly related to entropy. In other words, it may be hypothesized that entropy underlies the musical dwell-time effect rather than boundary status per se. The current experiment thus investigates the contributions of boundary status and predictive uncertainty to the musical dwell time effect by controlling these usually highly-correlated factors independently. In this procedure, participants selfpace through short melodies (derived from a corpus of Bach chorales) using a computer key to control the onset of each successive note with the explicit goal of optimizing their memorization of the music. Each melody contains a target note that (1) is phrase ending or beginning and (2) has high or low entropy (as estimated by the Information Dynamics of Music Model, IDyOM, trained on a large corpus of hymns and folksongs). Data collection is ongoing. The main analysis will examine whether longer dwelling is associated with boundary status or entropy. Results from this study will extend recent work on predictive uncertainty to the timing domain, as well as potentially answer key questions relating to boundary perception in musical listening.