ImportanceParent-infant bonding contributes to long-term infant health but may be disrupted by preterm birth.ObjectiveTo determine if parent-led, infant-directed singing, supported by a music therapist and initiated in the neonatal intensive care unit (NICU), improves parent-infant bonding at 6 and 12 months.Design, Setting, and ParticipantsThis randomized clinical trial was conducted in level III and IV NICUs in 5 countries between 2018 and 2022. Eligible participants were preterm infants (under 35 weeks’ gestation) and their parents. Follow-up was conducted across 12 months (as part of the LongSTEP study) at home or in clinics. Final follow-up was conducted at 12 months’ infant-corrected age. Data were analyzed from August 2022 to November 2022.InterventionParticipants randomized to music therapy (MT) plus standard care or standard care alone during NICU admission, or to MT plus standard care or standard care alone postdischarge, using computer-generated randomization (ratio 1:1, block sizes of 2 or 4 varying randomly), stratified by site (51 allocated to MT NICU, 53 to MT postdischarge, 52 to both, and 50 to neither). MT consisted of parent-led, infant-directed singing tailored to infant responses and supported by a music therapist 3 times per week throughout hospitalization or 7 sessions across 6 months’ postdischarge.Main Outcome and MeasurePrimary outcome was mother-infant bonding at 6 months’ corrected age, measured by the Postpartum Bonding Questionnaire (PBQ), with follow-up at 12 months’ corrected age, and analyzed intention-to-treat as group differences.ResultsOf 206 enrolled infants with 206 mothers (mean [SD] age, 33 [6] years) and 194 fathers (mean [SD] age, 36 [6] years) randomized at discharge, 196 (95.1%) completed assessments at 6 months and were analyzed. Estimated group effects for PBQ at 6 months’ corrected age were 0.55 (95% CI, −2.20 to 3.30; P = .70) for MT in the NICU, 1.02 (95% CI, −1.72 to 3.76; P = .47) for MT postdischarge, and −0.20 (95% CI, −4.03 to 3.63; P = .92) for the interaction (12 months: MT in NICU, 0.17; 95% CI, −2.71 to 3.05; P = .91; MT postdischarge, 1.78; 95% CI, −1.13 to 4.70; P = .24; interaction, −1.68; 95% CI, −5.77 to 2.41; P = .42). There were no clinically important between-group differences for secondary variables.Conclusions and RelevanceIn this randomized clinical trial, parent-led, infant-directed singing did not have clinically important effects on mother-infant bonding, but was safe and well-accepted.Trial RegistrationClinicalTrials.gov Identifier: NCT03564184