Parents commonly vocalize to infants to mitigate their distress, especially when holding them is not possible. Here we examined the relative efficacy of parents' speech and singing (familiar and unfamiliar songs) in alleviating the distress of 8-and 10-month-old infants (n = 68 per age group). Parent-infant dyads participated in three trials of the Still Face procedure, featuring a 2-min Play phase, a Still Face phase (parents immobile and unresponsive for 1 min or until infants became visibly distressed), and a 2-min Reunion phase in which caregivers attempted to reverse infant distress by 1) singing a highly familiar song, 2) singing an unfamiliar song, or 3) expressive talking (order counterbalanced across dyads). In the Reunion phase, talking led to increased negative affect in both age groups, in contrast to singing familiar or unfamiliar songs, which increased infant attention to parent and decreased negative affect. The favorable consequences were greatest for familiar songs, which also generated increased smiling. Skin conductance recorded from a subset of infants (n = 36 younger, 41 older infants) revealed that arousal levels were highest for the talking reunion, lowest for unfamiliar songs, and intermediate for familiar songs. The arousal effects, considered in conjunction with the behavioral effects, confirm that songs are more effective than speech at mitigating infant distress. We suggest, moreover, that familiar songs generate higher infant arousal than unfamiliar songs because they evoke excitement, reflected in modestly elevated arousal as well as pleasure, in contrast to more subdued responses to unfamiliar songs.