Much is known about the efficacy of infant-directed (ID) speech and singing for capturing attention, but little is known about their role in regulating affect. In Experiment 1, infants 7–10 months of age listened to scripted recordings of ID speech, adult-directed speech, or singing in an unfamiliar language (Turkish) until they met a criterion of distress based on negative facial expression. They listened to singing for roughly twice as long as speech before meeting the distress criterion. In Experiment 2, they were exposed to natural recordings of ID speech or singing in a familiar language. As in Experiment 1, ID singing was considerably more effective than speech for delaying the onset of distress. We suggest that the temporal patterning of ID singing, with its regular beat, metrical organization, and tempo, plays an important role in inhibiting distress, perhaps by promoting entrainment and predictive listening.